Mixed-precision quantizations of Latitude Games's 12B models, using NVFP4+4/6 for MLP layers and FP8_DYNAMIC for self-attention.
Jesse Mitchell
DataSnake
AI & ML interests
None yet
Organizations
None yet
models 18
DataSnake/Muse-12B-NVFP4-FP8
Text Generation • 8B • Updated • 48
DataSnake/Wayfarer-12B-NVFP4-FP8
Text Generation • 8B • Updated • 47
DataSnake/Mistral-Nemo-Instruct-2407-NVFP4-FP8
Text Generation • 8B • Updated • 244
DataSnake/Mistral-Nemo-Instruct-2407-NVFP4-FP8-RTN
8B • Updated • 12
DataSnake/Mistral-Nemo-Instruct-2407-Down-4over6
Text Generation • 9B • Updated • 6
DataSnake/Wayfarer-12B-NVFP4-4over6
Text Generation • 7B • Updated • 10 • 1
DataSnake/Mistral-Nemo-Instruct-2407-NVFP4-4over6
Text Generation • 7B • Updated • 12
DataSnake/Wayfarer-2-12B-NVFP4-FP8
Text Generation • 8B • Updated • 24
DataSnake/Wayfarer-2-12B-NVFP4-4over6
Text Generation • 7B • Updated • 4 • 2
DataSnake/Muse-12B-NVFP4-4over6
Text Generation • 7B • Updated • 46 • 1
datasets 0
None public yet