Instructions to use mlx-community/pixtral-12b-8bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mlx-community/pixtral-12b-8bit with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="mlx-community/pixtral-12b-8bit")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("mlx-community/pixtral-12b-8bit")
model = AutoModelForImageTextToText.from_pretrained("mlx-community/pixtral-12b-8bit")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

MLX

How to use mlx-community/pixtral-12b-8bit with MLX:

# Make sure mlx-vlm is installed
# pip install --upgrade mlx-vlm

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("mlx-community/pixtral-12b-8bit")
config = load_config("mlx-community/pixtral-12b-8bit")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps
LM Studio

vLLM

How to use mlx-community/pixtral-12b-8bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mlx-community/pixtral-12b-8bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/pixtral-12b-8bit",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/mlx-community/pixtral-12b-8bit

SGLang

How to use mlx-community/pixtral-12b-8bit with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "mlx-community/pixtral-12b-8bit" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/pixtral-12b-8bit",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "mlx-community/pixtral-12b-8bit" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mlx-community/pixtral-12b-8bit",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use mlx-community/pixtral-12b-8bit with Docker Model Runner:
```
docker model run hf.co/mlx-community/pixtral-12b-8bit
```

raise ValueError(f"Received parameters not in model: {extras}.")

by VectorNomad - opened Oct 2, 2024

Discussion

VectorNomad

MLX Community org Oct 2, 2024

I am trying to load the model using latest from mlx-vlm but getting this error

ValueError: Received parameters not in model: vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.17.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.5.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.1.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.1.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.7.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.12.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.11.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.0.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.22.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.4.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.16.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.18.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.8.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.4.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.21.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.18.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.12.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.ln_pre.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.0.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.23.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.19.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.2.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.4.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.23.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.6.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.22.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.14.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.11.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.4.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.8.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.patch_conv.weight vision_tower.vision_model.vision_model.transformer.layers.6.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.21.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.12.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.9.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.1.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.0.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.16.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.20.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.10.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.1.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.20.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.0.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.12.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.2.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.1.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.6.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.18.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.12.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.18.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.4.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.12.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.20.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.4.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.10.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.20.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.1.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.18.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.1.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.2.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.5.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.12.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.10.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.11.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.2.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.18.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.6.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.11.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.23.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.7.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.6.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.12.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.0.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.10.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.5.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.18.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.6.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.7.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.7.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.10.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.23.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.16.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.17.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.10.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.16.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.12.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.19.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.4.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.4.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.17.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.23.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.20.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.17.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.6.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.23.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.2.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.6.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.20.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.9.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.13.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.18.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.18.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.5.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.6.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.16.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.3.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.21.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.7.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.8.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.14.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.1.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.9.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.5.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.20.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.14.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.1.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.17.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.13.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.0.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.21.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention.o_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.14.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.16.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.9.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.5.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.18.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.8.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.13.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.21.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.2.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.23.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.16.feed_forward.down_proj.weight vision_tower.vision_model.vision_model.transformer.layers.4.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.17.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.down_proj.biases vision_tower.vision_model.vision_model.transformer.layers.0.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.9.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.22.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.attention.k_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.6.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.1.feed_forward.up_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.15.attention.v_proj.scales vision_tower.vision_model.vision_model.transformer.layers.20.attention.k_proj.biases vision_tower.vision_model.vision_model.transformer.layers.3.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.7.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.2.feed_forward.down_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention.q_proj.scales vision_tower.vision_model.vision_model.transformer.layers.18.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.14.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.23.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.15.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.3.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.23.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.7.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.15.feed_forward.up_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.feed_forward.gate_proj.weight vision_tower.vision_model.vision_model.transformer.layers.3.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.10.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.20.feed_forward.up_proj.weight vision_tower.vision_model.vision_model.transformer.layers.12.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.7.feed_forward.gate_proj.biases vision_tower.vision_model.vision_model.transformer.layers.19.attention.o_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.13.attention.k_proj.scales vision_tower.vision_model.vision_model.transformer.layers.0.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.6.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.14.feed_forward.gate_proj.scales vision_tower.vision_model.vision_model.transformer.layers.10.attention_norm.weight vision_tower.vision_model.vision_model.transformer.layers.4.ffn_norm.weight vision_tower.vision_model.vision_model.transformer.layers.22.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.10.attention.o_proj.scales vision_tower.vision_model.vision_model.transformer.layers.15.attention.q_proj.weight vision_tower.vision_model.vision_model.transformer.layers.21.attention.q_proj.biases vision_tower.vision_model.vision_model.transformer.layers.5.attention.v_proj.weight vision_tower.vision_model.vision_model.transformer.layers.11.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.23.attention.v_proj.biases vision_tower.vision_model.vision_model.transformer.layers.4.attention.o_proj.scales.

mc0ps

MLX Community org Oct 3, 2024

there was a bug in the quantization for these models. but you can download and convert the source models from the mistral repo and it should work fine:

https://huggingface.co/mistral-community/pixtral-12b/tree/main