Instructions to use SVPG-Labs/deff-1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SVPG-Labs/deff-1 with Transformers:

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

pipe = pipeline("image-text-to-text", model="SVPG-Labs/deff-1")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("SVPG-Labs/deff-1")
model = AutoModelForMultimodalLM.from_pretrained("SVPG-Labs/deff-1", device_map="auto")
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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use SVPG-Labs/deff-1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SVPG-Labs/deff-1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SVPG-Labs/deff-1",
		"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/SVPG-Labs/deff-1

SGLang

How to use SVPG-Labs/deff-1 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 "SVPG-Labs/deff-1" \
    --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": "SVPG-Labs/deff-1",
		"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 "SVPG-Labs/deff-1" \
        --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": "SVPG-Labs/deff-1",
		"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 SVPG-Labs/deff-1 with Docker Model Runner:
```
docker model run hf.co/SVPG-Labs/deff-1
```

deff-1 / quantization /hf_quant_config.modelopt-export.json

totallytim

Upload Deff-1 weights

01badea verified 5 days ago

Raw History Blame Contribute Delete

25.2 kB

{
    "producer": {
        "name": "modelopt",
        "version": "0.0.1.dev1+gd0142c9dc.d20261002"
    },
    "quantization": {
        "quant_algo": "NVFP4",
        "kv_cache_quant_algo": "FP8",
        "group_size": 16,
        "exclude_modules": [
            "lm_head",
            "model.decoder.embed_tokens",
            "model.decoder.layers.0.experts",
            "model.decoder.layers.0.input_layernorm",
            "model.decoder.layers.0.mlp*",
            "model.decoder.layers.0.post_attention_layernorm",
            "model.decoder.layers.0.post_feedforward_layernorm",
            "model.decoder.layers.0.post_feedforward_layernorm_1",
            "model.decoder.layers.0.post_feedforward_layernorm_2",
            "model.decoder.layers.0.pre_feedforward_layernorm",
            "model.decoder.layers.0.pre_feedforward_layernorm_2",
            "model.decoder.layers.0.router*",
            "model.decoder.layers.0.self_attn*",
            "model.decoder.layers.1.experts",
            "model.decoder.layers.1.input_layernorm",
            "model.decoder.layers.1.mlp*",
            "model.decoder.layers.1.post_attention_layernorm",
            "model.decoder.layers.1.post_feedforward_layernorm",
            "model.decoder.layers.1.post_feedforward_layernorm_1",
            "model.decoder.layers.1.post_feedforward_layernorm_2",
            "model.decoder.layers.1.pre_feedforward_layernorm",
            "model.decoder.layers.1.pre_feedforward_layernorm_2",
            "model.decoder.layers.1.router*",
            "model.decoder.layers.1.self_attn*",
            "model.decoder.layers.10.experts",
            "model.decoder.layers.10.input_layernorm",
            "model.decoder.layers.10.mlp*",
            "model.decoder.layers.10.post_attention_layernorm",
            "model.decoder.layers.10.post_feedforward_layernorm",
            "model.decoder.layers.10.post_feedforward_layernorm_1",
            "model.decoder.layers.10.post_feedforward_layernorm_2",
            "model.decoder.layers.10.pre_feedforward_layernorm",
            "model.decoder.layers.10.pre_feedforward_layernorm_2",
            "model.decoder.layers.10.router*",
            "model.decoder.layers.10.self_attn*",
            "model.decoder.layers.11.experts",
            "model.decoder.layers.11.input_layernorm",
            "model.decoder.layers.11.mlp*",
            "model.decoder.layers.11.post_attention_layernorm",
            "model.decoder.layers.11.post_feedforward_layernorm",
            "model.decoder.layers.11.post_feedforward_layernorm_1",
            "model.decoder.layers.11.post_feedforward_layernorm_2",
            "model.decoder.layers.11.pre_feedforward_layernorm",
            "model.decoder.layers.11.pre_feedforward_layernorm_2",
            "model.decoder.layers.11.router*",
            "model.decoder.layers.11.self_attn*",
            "model.decoder.layers.12.experts",
            "model.decoder.layers.12.input_layernorm",
            "model.decoder.layers.12.mlp*",
            "model.decoder.layers.12.post_attention_layernorm",
            "model.decoder.layers.12.post_feedforward_layernorm",
            "model.decoder.layers.12.post_feedforward_layernorm_1",
            "model.decoder.layers.12.post_feedforward_layernorm_2",
            "model.decoder.layers.12.pre_feedforward_layernorm",
            "model.decoder.layers.12.pre_feedforward_layernorm_2",
            "model.decoder.layers.12.router*",
            "model.decoder.layers.12.self_attn*",
            "model.decoder.layers.13.experts",
            "model.decoder.layers.13.input_layernorm",
            "model.decoder.layers.13.mlp*",
            "model.decoder.layers.13.post_attention_layernorm",
            "model.decoder.layers.13.post_feedforward_layernorm",
            "model.decoder.layers.13.post_feedforward_layernorm_1",
            "model.decoder.layers.13.post_feedforward_layernorm_2",
            "model.decoder.layers.13.pre_feedforward_layernorm",
            "model.decoder.layers.13.pre_feedforward_layernorm_2",
            "model.decoder.layers.13.router*",
            "model.decoder.layers.13.self_attn*",
            "model.decoder.layers.14.experts",
            "model.decoder.layers.14.input_layernorm",
            "model.decoder.layers.14.mlp*",
            "model.decoder.layers.14.post_attention_layernorm",
            "model.decoder.layers.14.post_feedforward_layernorm",
            "model.decoder.layers.14.post_feedforward_layernorm_1",
            "model.decoder.layers.14.post_feedforward_layernorm_2",
            "model.decoder.layers.14.pre_feedforward_layernorm",
            "model.decoder.layers.14.pre_feedforward_layernorm_2",
            "model.decoder.layers.14.router*",
            "model.decoder.layers.14.self_attn*",
            "model.decoder.layers.15.experts",
            "model.decoder.layers.15.input_layernorm",
            "model.decoder.layers.15.mlp*",
            "model.decoder.layers.15.post_attention_layernorm",
            "model.decoder.layers.15.post_feedforward_layernorm",
            "model.decoder.layers.15.post_feedforward_layernorm_1",
            "model.decoder.layers.15.post_feedforward_layernorm_2",
            "model.decoder.layers.15.pre_feedforward_layernorm",
            "model.decoder.layers.15.pre_feedforward_layernorm_2",
            "model.decoder.layers.15.router*",
            "model.decoder.layers.15.self_attn*",
            "model.decoder.layers.16.experts",
            "model.decoder.layers.16.input_layernorm",
            "model.decoder.layers.16.mlp*",
            "model.decoder.layers.16.post_attention_layernorm",
            "model.decoder.layers.16.post_feedforward_layernorm",
            "model.decoder.layers.16.post_feedforward_layernorm_1",
            "model.decoder.layers.16.post_feedforward_layernorm_2",
            "model.decoder.layers.16.pre_feedforward_layernorm",
            "model.decoder.layers.16.pre_feedforward_layernorm_2",
            "model.decoder.layers.16.router*",
            "model.decoder.layers.16.self_attn*",
            "model.decoder.layers.17.experts",
            "model.decoder.layers.17.input_layernorm",
            "model.decoder.layers.17.mlp*",
            "model.decoder.layers.17.post_attention_layernorm",
            "model.decoder.layers.17.post_feedforward_layernorm",
            "model.decoder.layers.17.post_feedforward_layernorm_1",
            "model.decoder.layers.17.post_feedforward_layernorm_2",
            "model.decoder.layers.17.pre_feedforward_layernorm",
            "model.decoder.layers.17.pre_feedforward_layernorm_2",
            "model.decoder.layers.17.router*",
            "model.decoder.layers.17.self_attn*",
            "model.decoder.layers.18.experts",
            "model.decoder.layers.18.input_layernorm",
            "model.decoder.layers.18.mlp*",
            "model.decoder.layers.18.post_attention_layernorm",
            "model.decoder.layers.18.post_feedforward_layernorm",
            "model.decoder.layers.18.post_feedforward_layernorm_1",
            "model.decoder.layers.18.post_feedforward_layernorm_2",
            "model.decoder.layers.18.pre_feedforward_layernorm",
            "model.decoder.layers.18.pre_feedforward_layernorm_2",
            "model.decoder.layers.18.router*",
            "model.decoder.layers.18.self_attn*",
            "model.decoder.layers.19.experts",
            "model.decoder.layers.19.input_layernorm",
            "model.decoder.layers.19.mlp*",
            "model.decoder.layers.19.post_attention_layernorm",
            "model.decoder.layers.19.post_feedforward_layernorm",
            "model.decoder.layers.19.post_feedforward_layernorm_1",
            "model.decoder.layers.19.post_feedforward_layernorm_2",
            "model.decoder.layers.19.pre_feedforward_layernorm",
            "model.decoder.layers.19.pre_feedforward_layernorm_2",
            "model.decoder.layers.19.router*",
            "model.decoder.layers.19.self_attn*",
            "model.decoder.layers.2.experts",
            "model.decoder.layers.2.input_layernorm",
            "model.decoder.layers.2.mlp*",
            "model.decoder.layers.2.post_attention_layernorm",
            "model.decoder.layers.2.post_feedforward_layernorm",
            "model.decoder.layers.2.post_feedforward_layernorm_1",
            "model.decoder.layers.2.post_feedforward_layernorm_2",
            "model.decoder.layers.2.pre_feedforward_layernorm",
            "model.decoder.layers.2.pre_feedforward_layernorm_2",
            "model.decoder.layers.2.router*",
            "model.decoder.layers.2.self_attn*",
            "model.decoder.layers.20.experts",
            "model.decoder.layers.20.input_layernorm",
            "model.decoder.layers.20.mlp*",
            "model.decoder.layers.20.post_attention_layernorm",
            "model.decoder.layers.20.post_feedforward_layernorm",
            "model.decoder.layers.20.post_feedforward_layernorm_1",
            "model.decoder.layers.20.post_feedforward_layernorm_2",
            "model.decoder.layers.20.pre_feedforward_layernorm",
            "model.decoder.layers.20.pre_feedforward_layernorm_2",
            "model.decoder.layers.20.router*",
            "model.decoder.layers.20.self_attn*",
            "model.decoder.layers.21.experts",
            "model.decoder.layers.21.input_layernorm",
            "model.decoder.layers.21.mlp*",
            "model.decoder.layers.21.post_attention_layernorm",
            "model.decoder.layers.21.post_feedforward_layernorm",
            "model.decoder.layers.21.post_feedforward_layernorm_1",
            "model.decoder.layers.21.post_feedforward_layernorm_2",
            "model.decoder.layers.21.pre_feedforward_layernorm",
            "model.decoder.layers.21.pre_feedforward_layernorm_2",
            "model.decoder.layers.21.router*",
            "model.decoder.layers.21.self_attn*",
            "model.decoder.layers.22.experts",
            "model.decoder.layers.22.input_layernorm",
            "model.decoder.layers.22.mlp*",
            "model.decoder.layers.22.post_attention_layernorm",
            "model.decoder.layers.22.post_feedforward_layernorm",
            "model.decoder.layers.22.post_feedforward_layernorm_1",
            "model.decoder.layers.22.post_feedforward_layernorm_2",
            "model.decoder.layers.22.pre_feedforward_layernorm",
            "model.decoder.layers.22.pre_feedforward_layernorm_2",
            "model.decoder.layers.22.router*",
            "model.decoder.layers.22.self_attn*",
            "model.decoder.layers.23.experts",
            "model.decoder.layers.23.input_layernorm",
            "model.decoder.layers.23.mlp*",
            "model.decoder.layers.23.post_attention_layernorm",
            "model.decoder.layers.23.post_feedforward_layernorm",
            "model.decoder.layers.23.post_feedforward_layernorm_1",
            "model.decoder.layers.23.post_feedforward_layernorm_2",
            "model.decoder.layers.23.pre_feedforward_layernorm",
            "model.decoder.layers.23.pre_feedforward_layernorm_2",
            "model.decoder.layers.23.router*",
            "model.decoder.layers.23.self_attn*",
            "model.decoder.layers.24.experts",
            "model.decoder.layers.24.input_layernorm",
            "model.decoder.layers.24.mlp*",
            "model.decoder.layers.24.post_attention_layernorm",
            "model.decoder.layers.24.post_feedforward_layernorm",
            "model.decoder.layers.24.post_feedforward_layernorm_1",
            "model.decoder.layers.24.post_feedforward_layernorm_2",
            "model.decoder.layers.24.pre_feedforward_layernorm",
            "model.decoder.layers.24.pre_feedforward_layernorm_2",
            "model.decoder.layers.24.router*",
            "model.decoder.layers.24.self_attn*",
            "model.decoder.layers.25.experts",
            "model.decoder.layers.25.input_layernorm",
            "model.decoder.layers.25.mlp*",
            "model.decoder.layers.25.post_attention_layernorm",
            "model.decoder.layers.25.post_feedforward_layernorm",
            "model.decoder.layers.25.post_feedforward_layernorm_1",
            "model.decoder.layers.25.post_feedforward_layernorm_2",
            "model.decoder.layers.25.pre_feedforward_layernorm",
            "model.decoder.layers.25.pre_feedforward_layernorm_2",
            "model.decoder.layers.25.router*",
            "model.decoder.layers.25.self_attn*",
            "model.decoder.layers.26.experts",
            "model.decoder.layers.26.input_layernorm",
            "model.decoder.layers.26.mlp*",
            "model.decoder.layers.26.post_attention_layernorm",
            "model.decoder.layers.26.post_feedforward_layernorm",
            "model.decoder.layers.26.post_feedforward_layernorm_1",
            "model.decoder.layers.26.post_feedforward_layernorm_2",
            "model.decoder.layers.26.pre_feedforward_layernorm",
            "model.decoder.layers.26.pre_feedforward_layernorm_2",
            "model.decoder.layers.26.router*",
            "model.decoder.layers.26.self_attn*",
            "model.decoder.layers.27.experts",
            "model.decoder.layers.27.input_layernorm",
            "model.decoder.layers.27.mlp*",
            "model.decoder.layers.27.post_attention_layernorm",
            "model.decoder.layers.27.post_feedforward_layernorm",
            "model.decoder.layers.27.post_feedforward_layernorm_1",
            "model.decoder.layers.27.post_feedforward_layernorm_2",
            "model.decoder.layers.27.pre_feedforward_layernorm",
            "model.decoder.layers.27.pre_feedforward_layernorm_2",
            "model.decoder.layers.27.router*",
            "model.decoder.layers.27.self_attn*",
            "model.decoder.layers.28.experts",
            "model.decoder.layers.28.input_layernorm",
            "model.decoder.layers.28.mlp*",
            "model.decoder.layers.28.post_attention_layernorm",
            "model.decoder.layers.28.post_feedforward_layernorm",
            "model.decoder.layers.28.post_feedforward_layernorm_1",
            "model.decoder.layers.28.post_feedforward_layernorm_2",
            "model.decoder.layers.28.pre_feedforward_layernorm",
            "model.decoder.layers.28.pre_feedforward_layernorm_2",
            "model.decoder.layers.28.router*",
            "model.decoder.layers.28.self_attn*",
            "model.decoder.layers.29.experts",
            "model.decoder.layers.29.input_layernorm",
            "model.decoder.layers.29.mlp*",
            "model.decoder.layers.29.post_attention_layernorm",
            "model.decoder.layers.29.post_feedforward_layernorm",
            "model.decoder.layers.29.post_feedforward_layernorm_1",
            "model.decoder.layers.29.post_feedforward_layernorm_2",
            "model.decoder.layers.29.pre_feedforward_layernorm",
            "model.decoder.layers.29.pre_feedforward_layernorm_2",
            "model.decoder.layers.29.router*",
            "model.decoder.layers.29.self_attn*",
            "model.decoder.layers.3.experts",
            "model.decoder.layers.3.input_layernorm",
            "model.decoder.layers.3.mlp*",
            "model.decoder.layers.3.post_attention_layernorm",
            "model.decoder.layers.3.post_feedforward_layernorm",
            "model.decoder.layers.3.post_feedforward_layernorm_1",
            "model.decoder.layers.3.post_feedforward_layernorm_2",
            "model.decoder.layers.3.pre_feedforward_layernorm",
            "model.decoder.layers.3.pre_feedforward_layernorm_2",
            "model.decoder.layers.3.router*",
            "model.decoder.layers.3.self_attn*",
            "model.decoder.layers.4.experts",
            "model.decoder.layers.4.input_layernorm",
            "model.decoder.layers.4.mlp*",
            "model.decoder.layers.4.post_attention_layernorm",
            "model.decoder.layers.4.post_feedforward_layernorm",
            "model.decoder.layers.4.post_feedforward_layernorm_1",
            "model.decoder.layers.4.post_feedforward_layernorm_2",
            "model.decoder.layers.4.pre_feedforward_layernorm",
            "model.decoder.layers.4.pre_feedforward_layernorm_2",
            "model.decoder.layers.4.router*",
            "model.decoder.layers.4.self_attn*",
            "model.decoder.layers.5.experts",
            "model.decoder.layers.5.input_layernorm",
            "model.decoder.layers.5.mlp*",
            "model.decoder.layers.5.post_attention_layernorm",
            "model.decoder.layers.5.post_feedforward_layernorm",
            "model.decoder.layers.5.post_feedforward_layernorm_1",
            "model.decoder.layers.5.post_feedforward_layernorm_2",
            "model.decoder.layers.5.pre_feedforward_layernorm",
            "model.decoder.layers.5.pre_feedforward_layernorm_2",
            "model.decoder.layers.5.router*",
            "model.decoder.layers.5.self_attn*",
            "model.decoder.layers.6.experts",
            "model.decoder.layers.6.input_layernorm",
            "model.decoder.layers.6.mlp*",
            "model.decoder.layers.6.post_attention_layernorm",
            "model.decoder.layers.6.post_feedforward_layernorm",
            "model.decoder.layers.6.post_feedforward_layernorm_1",
            "model.decoder.layers.6.post_feedforward_layernorm_2",
            "model.decoder.layers.6.pre_feedforward_layernorm",
            "model.decoder.layers.6.pre_feedforward_layernorm_2",
            "model.decoder.layers.6.router*",
            "model.decoder.layers.6.self_attn*",
            "model.decoder.layers.7.experts",
            "model.decoder.layers.7.input_layernorm",
            "model.decoder.layers.7.mlp*",
            "model.decoder.layers.7.post_attention_layernorm",
            "model.decoder.layers.7.post_feedforward_layernorm",
            "model.decoder.layers.7.post_feedforward_layernorm_1",
            "model.decoder.layers.7.post_feedforward_layernorm_2",
            "model.decoder.layers.7.pre_feedforward_layernorm",
            "model.decoder.layers.7.pre_feedforward_layernorm_2",
            "model.decoder.layers.7.router*",
            "model.decoder.layers.7.self_attn*",
            "model.decoder.layers.8.experts",
            "model.decoder.layers.8.input_layernorm",
            "model.decoder.layers.8.mlp*",
            "model.decoder.layers.8.post_attention_layernorm",
            "model.decoder.layers.8.post_feedforward_layernorm",
            "model.decoder.layers.8.post_feedforward_layernorm_1",
            "model.decoder.layers.8.post_feedforward_layernorm_2",
            "model.decoder.layers.8.pre_feedforward_layernorm",
            "model.decoder.layers.8.pre_feedforward_layernorm_2",
            "model.decoder.layers.8.router*",
            "model.decoder.layers.8.self_attn*",
            "model.decoder.layers.9.experts",
            "model.decoder.layers.9.input_layernorm",
            "model.decoder.layers.9.mlp*",
            "model.decoder.layers.9.post_attention_layernorm",
            "model.decoder.layers.9.post_feedforward_layernorm",
            "model.decoder.layers.9.post_feedforward_layernorm_1",
            "model.decoder.layers.9.post_feedforward_layernorm_2",
            "model.decoder.layers.9.pre_feedforward_layernorm",
            "model.decoder.layers.9.pre_feedforward_layernorm_2",
            "model.decoder.layers.9.router*",
            "model.decoder.layers.9.self_attn*",
            "model.decoder.norm",
            "model.decoder.self_conditioning*",
            "model.encoder.embed_vision*",
            "model.encoder.language_model.layers.0.mlp*",
            "model.encoder.language_model.layers.0.router*",
            "model.encoder.language_model.layers.0.self_attn*",
            "model.encoder.language_model.layers.1.mlp*",
            "model.encoder.language_model.layers.1.router*",
            "model.encoder.language_model.layers.1.self_attn*",
            "model.encoder.language_model.layers.10.mlp*",
            "model.encoder.language_model.layers.10.router*",
            "model.encoder.language_model.layers.10.self_attn*",
            "model.encoder.language_model.layers.11.mlp*",
            "model.encoder.language_model.layers.11.router*",
            "model.encoder.language_model.layers.11.self_attn*",
            "model.encoder.language_model.layers.12.mlp*",
            "model.encoder.language_model.layers.12.router*",
            "model.encoder.language_model.layers.12.self_attn*",
            "model.encoder.language_model.layers.13.mlp*",
            "model.encoder.language_model.layers.13.router*",
            "model.encoder.language_model.layers.13.self_attn*",
            "model.encoder.language_model.layers.14.mlp*",
            "model.encoder.language_model.layers.14.router*",
            "model.encoder.language_model.layers.14.self_attn*",
            "model.encoder.language_model.layers.15.mlp*",
            "model.encoder.language_model.layers.15.router*",
            "model.encoder.language_model.layers.15.self_attn*",
            "model.encoder.language_model.layers.16.mlp*",
            "model.encoder.language_model.layers.16.router*",
            "model.encoder.language_model.layers.16.self_attn*",
            "model.encoder.language_model.layers.17.mlp*",
            "model.encoder.language_model.layers.17.router*",
            "model.encoder.language_model.layers.17.self_attn*",
            "model.encoder.language_model.layers.18.mlp*",
            "model.encoder.language_model.layers.18.router*",
            "model.encoder.language_model.layers.18.self_attn*",
            "model.encoder.language_model.layers.19.mlp*",
            "model.encoder.language_model.layers.19.router*",
            "model.encoder.language_model.layers.19.self_attn*",
            "model.encoder.language_model.layers.2.mlp*",
            "model.encoder.language_model.layers.2.router*",
            "model.encoder.language_model.layers.2.self_attn*",
            "model.encoder.language_model.layers.20.mlp*",
            "model.encoder.language_model.layers.20.router*",
            "model.encoder.language_model.layers.20.self_attn*",
            "model.encoder.language_model.layers.21.mlp*",
            "model.encoder.language_model.layers.21.router*",
            "model.encoder.language_model.layers.21.self_attn*",
            "model.encoder.language_model.layers.22.mlp*",
            "model.encoder.language_model.layers.22.router*",
            "model.encoder.language_model.layers.22.self_attn*",
            "model.encoder.language_model.layers.23.mlp*",
            "model.encoder.language_model.layers.23.router*",
            "model.encoder.language_model.layers.23.self_attn*",
            "model.encoder.language_model.layers.24.mlp*",
            "model.encoder.language_model.layers.24.router*",
            "model.encoder.language_model.layers.24.self_attn*",
            "model.encoder.language_model.layers.25.mlp*",
            "model.encoder.language_model.layers.25.router*",
            "model.encoder.language_model.layers.25.self_attn*",
            "model.encoder.language_model.layers.26.mlp*",
            "model.encoder.language_model.layers.26.router*",
            "model.encoder.language_model.layers.26.self_attn*",
            "model.encoder.language_model.layers.27.mlp*",
            "model.encoder.language_model.layers.27.router*",
            "model.encoder.language_model.layers.27.self_attn*",
            "model.encoder.language_model.layers.28.mlp*",
            "model.encoder.language_model.layers.28.router*",
            "model.encoder.language_model.layers.28.self_attn*",
            "model.encoder.language_model.layers.29.mlp*",
            "model.encoder.language_model.layers.29.router*",
            "model.encoder.language_model.layers.29.self_attn*",
            "model.encoder.language_model.layers.3.mlp*",
            "model.encoder.language_model.layers.3.router*",
            "model.encoder.language_model.layers.3.self_attn*",
            "model.encoder.language_model.layers.4.mlp*",
            "model.encoder.language_model.layers.4.router*",
            "model.encoder.language_model.layers.4.self_attn*",
            "model.encoder.language_model.layers.5.mlp*",
            "model.encoder.language_model.layers.5.router*",
            "model.encoder.language_model.layers.5.self_attn*",
            "model.encoder.language_model.layers.6.mlp*",
            "model.encoder.language_model.layers.6.router*",
            "model.encoder.language_model.layers.6.self_attn*",
            "model.encoder.language_model.layers.7.mlp*",
            "model.encoder.language_model.layers.7.router*",
            "model.encoder.language_model.layers.7.self_attn*",
            "model.encoder.language_model.layers.8.mlp*",
            "model.encoder.language_model.layers.8.router*",
            "model.encoder.language_model.layers.8.self_attn*",
            "model.encoder.language_model.layers.9.mlp*",
            "model.encoder.language_model.layers.9.router*",
            "model.encoder.language_model.layers.9.self_attn*",
            "model.encoder.vision_tower*"
        ]
    }
}

	{
	"producer": {
	"name": "modelopt",
	"version": "0.0.1.dev1+gd0142c9dc.d20261002"
	},
	"quantization": {
	"quant_algo": "NVFP4",
	"kv_cache_quant_algo": "FP8",
	"group_size": 16,
	"exclude_modules": [
	"lm_head",
	"model.decoder.embed_tokens",
	"model.decoder.layers.0.experts",
	"model.decoder.layers.0.input_layernorm",
	"model.decoder.layers.0.mlp*",
	"model.decoder.layers.0.post_attention_layernorm",
	"model.decoder.layers.0.post_feedforward_layernorm",
	"model.decoder.layers.0.post_feedforward_layernorm_1",
	"model.decoder.layers.0.post_feedforward_layernorm_2",
	"model.decoder.layers.0.pre_feedforward_layernorm",
	"model.decoder.layers.0.pre_feedforward_layernorm_2",
	"model.decoder.layers.0.router*",
	"model.decoder.layers.0.self_attn*",
	"model.decoder.layers.1.experts",
	"model.decoder.layers.1.input_layernorm",
	"model.decoder.layers.1.mlp*",
	"model.decoder.layers.1.post_attention_layernorm",
	"model.decoder.layers.1.post_feedforward_layernorm",
	"model.decoder.layers.1.post_feedforward_layernorm_1",
	"model.decoder.layers.1.post_feedforward_layernorm_2",
	"model.decoder.layers.1.pre_feedforward_layernorm",
	"model.decoder.layers.1.pre_feedforward_layernorm_2",
	"model.decoder.layers.1.router*",
	"model.decoder.layers.1.self_attn*",
	"model.decoder.layers.10.experts",
	"model.decoder.layers.10.input_layernorm",
	"model.decoder.layers.10.mlp*",
	"model.decoder.layers.10.post_attention_layernorm",
	"model.decoder.layers.10.post_feedforward_layernorm",
	"model.decoder.layers.10.post_feedforward_layernorm_1",
	"model.decoder.layers.10.post_feedforward_layernorm_2",
	"model.decoder.layers.10.pre_feedforward_layernorm",
	"model.decoder.layers.10.pre_feedforward_layernorm_2",
	"model.decoder.layers.10.router*",
	"model.decoder.layers.10.self_attn*",
	"model.decoder.layers.11.experts",
	"model.decoder.layers.11.input_layernorm",
	"model.decoder.layers.11.mlp*",
	"model.decoder.layers.11.post_attention_layernorm",
	"model.decoder.layers.11.post_feedforward_layernorm",
	"model.decoder.layers.11.post_feedforward_layernorm_1",
	"model.decoder.layers.11.post_feedforward_layernorm_2",
	"model.decoder.layers.11.pre_feedforward_layernorm",
	"model.decoder.layers.11.pre_feedforward_layernorm_2",
	"model.decoder.layers.11.router*",
	"model.decoder.layers.11.self_attn*",
	"model.decoder.layers.12.experts",
	"model.decoder.layers.12.input_layernorm",
	"model.decoder.layers.12.mlp*",
	"model.decoder.layers.12.post_attention_layernorm",
	"model.decoder.layers.12.post_feedforward_layernorm",
	"model.decoder.layers.12.post_feedforward_layernorm_1",
	"model.decoder.layers.12.post_feedforward_layernorm_2",
	"model.decoder.layers.12.pre_feedforward_layernorm",
	"model.decoder.layers.12.pre_feedforward_layernorm_2",
	"model.decoder.layers.12.router*",
	"model.decoder.layers.12.self_attn*",
	"model.decoder.layers.13.experts",
	"model.decoder.layers.13.input_layernorm",
	"model.decoder.layers.13.mlp*",
	"model.decoder.layers.13.post_attention_layernorm",
	"model.decoder.layers.13.post_feedforward_layernorm",
	"model.decoder.layers.13.post_feedforward_layernorm_1",
	"model.decoder.layers.13.post_feedforward_layernorm_2",
	"model.decoder.layers.13.pre_feedforward_layernorm",
	"model.decoder.layers.13.pre_feedforward_layernorm_2",
	"model.decoder.layers.13.router*",
	"model.decoder.layers.13.self_attn*",
	"model.decoder.layers.14.experts",
	"model.decoder.layers.14.input_layernorm",
	"model.decoder.layers.14.mlp*",
	"model.decoder.layers.14.post_attention_layernorm",
	"model.decoder.layers.14.post_feedforward_layernorm",
	"model.decoder.layers.14.post_feedforward_layernorm_1",
	"model.decoder.layers.14.post_feedforward_layernorm_2",
	"model.decoder.layers.14.pre_feedforward_layernorm",
	"model.decoder.layers.14.pre_feedforward_layernorm_2",
	"model.decoder.layers.14.router*",
	"model.decoder.layers.14.self_attn*",
	"model.decoder.layers.15.experts",
	"model.decoder.layers.15.input_layernorm",
	"model.decoder.layers.15.mlp*",
	"model.decoder.layers.15.post_attention_layernorm",
	"model.decoder.layers.15.post_feedforward_layernorm",
	"model.decoder.layers.15.post_feedforward_layernorm_1",
	"model.decoder.layers.15.post_feedforward_layernorm_2",
	"model.decoder.layers.15.pre_feedforward_layernorm",
	"model.decoder.layers.15.pre_feedforward_layernorm_2",
	"model.decoder.layers.15.router*",
	"model.decoder.layers.15.self_attn*",
	"model.decoder.layers.16.experts",
	"model.decoder.layers.16.input_layernorm",
	"model.decoder.layers.16.mlp*",
	"model.decoder.layers.16.post_attention_layernorm",
	"model.decoder.layers.16.post_feedforward_layernorm",
	"model.decoder.layers.16.post_feedforward_layernorm_1",
	"model.decoder.layers.16.post_feedforward_layernorm_2",
	"model.decoder.layers.16.pre_feedforward_layernorm",
	"model.decoder.layers.16.pre_feedforward_layernorm_2",
	"model.decoder.layers.16.router*",
	"model.decoder.layers.16.self_attn*",
	"model.decoder.layers.17.experts",
	"model.decoder.layers.17.input_layernorm",
	"model.decoder.layers.17.mlp*",
	"model.decoder.layers.17.post_attention_layernorm",
	"model.decoder.layers.17.post_feedforward_layernorm",
	"model.decoder.layers.17.post_feedforward_layernorm_1",
	"model.decoder.layers.17.post_feedforward_layernorm_2",
	"model.decoder.layers.17.pre_feedforward_layernorm",
	"model.decoder.layers.17.pre_feedforward_layernorm_2",
	"model.decoder.layers.17.router*",
	"model.decoder.layers.17.self_attn*",
	"model.decoder.layers.18.experts",
	"model.decoder.layers.18.input_layernorm",
	"model.decoder.layers.18.mlp*",
	"model.decoder.layers.18.post_attention_layernorm",
	"model.decoder.layers.18.post_feedforward_layernorm",
	"model.decoder.layers.18.post_feedforward_layernorm_1",
	"model.decoder.layers.18.post_feedforward_layernorm_2",
	"model.decoder.layers.18.pre_feedforward_layernorm",
	"model.decoder.layers.18.pre_feedforward_layernorm_2",
	"model.decoder.layers.18.router*",
	"model.decoder.layers.18.self_attn*",
	"model.decoder.layers.19.experts",
	"model.decoder.layers.19.input_layernorm",
	"model.decoder.layers.19.mlp*",
	"model.decoder.layers.19.post_attention_layernorm",
	"model.decoder.layers.19.post_feedforward_layernorm",
	"model.decoder.layers.19.post_feedforward_layernorm_1",
	"model.decoder.layers.19.post_feedforward_layernorm_2",
	"model.decoder.layers.19.pre_feedforward_layernorm",
	"model.decoder.layers.19.pre_feedforward_layernorm_2",
	"model.decoder.layers.19.router*",
	"model.decoder.layers.19.self_attn*",
	"model.decoder.layers.2.experts",
	"model.decoder.layers.2.input_layernorm",
	"model.decoder.layers.2.mlp*",
	"model.decoder.layers.2.post_attention_layernorm",
	"model.decoder.layers.2.post_feedforward_layernorm",
	"model.decoder.layers.2.post_feedforward_layernorm_1",
	"model.decoder.layers.2.post_feedforward_layernorm_2",
	"model.decoder.layers.2.pre_feedforward_layernorm",
	"model.decoder.layers.2.pre_feedforward_layernorm_2",
	"model.decoder.layers.2.router*",
	"model.decoder.layers.2.self_attn*",
	"model.decoder.layers.20.experts",
	"model.decoder.layers.20.input_layernorm",
	"model.decoder.layers.20.mlp*",
	"model.decoder.layers.20.post_attention_layernorm",
	"model.decoder.layers.20.post_feedforward_layernorm",
	"model.decoder.layers.20.post_feedforward_layernorm_1",
	"model.decoder.layers.20.post_feedforward_layernorm_2",
	"model.decoder.layers.20.pre_feedforward_layernorm",
	"model.decoder.layers.20.pre_feedforward_layernorm_2",
	"model.decoder.layers.20.router*",
	"model.decoder.layers.20.self_attn*",
	"model.decoder.layers.21.experts",
	"model.decoder.layers.21.input_layernorm",
	"model.decoder.layers.21.mlp*",
	"model.decoder.layers.21.post_attention_layernorm",
	"model.decoder.layers.21.post_feedforward_layernorm",
	"model.decoder.layers.21.post_feedforward_layernorm_1",
	"model.decoder.layers.21.post_feedforward_layernorm_2",
	"model.decoder.layers.21.pre_feedforward_layernorm",
	"model.decoder.layers.21.pre_feedforward_layernorm_2",
	"model.decoder.layers.21.router*",
	"model.decoder.layers.21.self_attn*",
	"model.decoder.layers.22.experts",
	"model.decoder.layers.22.input_layernorm",
	"model.decoder.layers.22.mlp*",
	"model.decoder.layers.22.post_attention_layernorm",
	"model.decoder.layers.22.post_feedforward_layernorm",
	"model.decoder.layers.22.post_feedforward_layernorm_1",
	"model.decoder.layers.22.post_feedforward_layernorm_2",
	"model.decoder.layers.22.pre_feedforward_layernorm",
	"model.decoder.layers.22.pre_feedforward_layernorm_2",
	"model.decoder.layers.22.router*",
	"model.decoder.layers.22.self_attn*",
	"model.decoder.layers.23.experts",
	"model.decoder.layers.23.input_layernorm",
	"model.decoder.layers.23.mlp*",
	"model.decoder.layers.23.post_attention_layernorm",
	"model.decoder.layers.23.post_feedforward_layernorm",
	"model.decoder.layers.23.post_feedforward_layernorm_1",
	"model.decoder.layers.23.post_feedforward_layernorm_2",
	"model.decoder.layers.23.pre_feedforward_layernorm",
	"model.decoder.layers.23.pre_feedforward_layernorm_2",
	"model.decoder.layers.23.router*",
	"model.decoder.layers.23.self_attn*",
	"model.decoder.layers.24.experts",
	"model.decoder.layers.24.input_layernorm",
	"model.decoder.layers.24.mlp*",
	"model.decoder.layers.24.post_attention_layernorm",
	"model.decoder.layers.24.post_feedforward_layernorm",
	"model.decoder.layers.24.post_feedforward_layernorm_1",
	"model.decoder.layers.24.post_feedforward_layernorm_2",
	"model.decoder.layers.24.pre_feedforward_layernorm",
	"model.decoder.layers.24.pre_feedforward_layernorm_2",
	"model.decoder.layers.24.router*",
	"model.decoder.layers.24.self_attn*",
	"model.decoder.layers.25.experts",
	"model.decoder.layers.25.input_layernorm",
	"model.decoder.layers.25.mlp*",
	"model.decoder.layers.25.post_attention_layernorm",
	"model.decoder.layers.25.post_feedforward_layernorm",
	"model.decoder.layers.25.post_feedforward_layernorm_1",
	"model.decoder.layers.25.post_feedforward_layernorm_2",
	"model.decoder.layers.25.pre_feedforward_layernorm",
	"model.decoder.layers.25.pre_feedforward_layernorm_2",
	"model.decoder.layers.25.router*",
	"model.decoder.layers.25.self_attn*",
	"model.decoder.layers.26.experts",
	"model.decoder.layers.26.input_layernorm",
	"model.decoder.layers.26.mlp*",
	"model.decoder.layers.26.post_attention_layernorm",
	"model.decoder.layers.26.post_feedforward_layernorm",
	"model.decoder.layers.26.post_feedforward_layernorm_1",
	"model.decoder.layers.26.post_feedforward_layernorm_2",
	"model.decoder.layers.26.pre_feedforward_layernorm",
	"model.decoder.layers.26.pre_feedforward_layernorm_2",
	"model.decoder.layers.26.router*",
	"model.decoder.layers.26.self_attn*",
	"model.decoder.layers.27.experts",
	"model.decoder.layers.27.input_layernorm",
	"model.decoder.layers.27.mlp*",
	"model.decoder.layers.27.post_attention_layernorm",
	"model.decoder.layers.27.post_feedforward_layernorm",
	"model.decoder.layers.27.post_feedforward_layernorm_1",
	"model.decoder.layers.27.post_feedforward_layernorm_2",
	"model.decoder.layers.27.pre_feedforward_layernorm",
	"model.decoder.layers.27.pre_feedforward_layernorm_2",
	"model.decoder.layers.27.router*",
	"model.decoder.layers.27.self_attn*",
	"model.decoder.layers.28.experts",
	"model.decoder.layers.28.input_layernorm",
	"model.decoder.layers.28.mlp*",
	"model.decoder.layers.28.post_attention_layernorm",
	"model.decoder.layers.28.post_feedforward_layernorm",
	"model.decoder.layers.28.post_feedforward_layernorm_1",
	"model.decoder.layers.28.post_feedforward_layernorm_2",
	"model.decoder.layers.28.pre_feedforward_layernorm",
	"model.decoder.layers.28.pre_feedforward_layernorm_2",
	"model.decoder.layers.28.router*",
	"model.decoder.layers.28.self_attn*",
	"model.decoder.layers.29.experts",
	"model.decoder.layers.29.input_layernorm",
	"model.decoder.layers.29.mlp*",
	"model.decoder.layers.29.post_attention_layernorm",
	"model.decoder.layers.29.post_feedforward_layernorm",
	"model.decoder.layers.29.post_feedforward_layernorm_1",
	"model.decoder.layers.29.post_feedforward_layernorm_2",
	"model.decoder.layers.29.pre_feedforward_layernorm",
	"model.decoder.layers.29.pre_feedforward_layernorm_2",
	"model.decoder.layers.29.router*",
	"model.decoder.layers.29.self_attn*",
	"model.decoder.layers.3.experts",
	"model.decoder.layers.3.input_layernorm",
	"model.decoder.layers.3.mlp*",
	"model.decoder.layers.3.post_attention_layernorm",
	"model.decoder.layers.3.post_feedforward_layernorm",
	"model.decoder.layers.3.post_feedforward_layernorm_1",
	"model.decoder.layers.3.post_feedforward_layernorm_2",
	"model.decoder.layers.3.pre_feedforward_layernorm",
	"model.decoder.layers.3.pre_feedforward_layernorm_2",
	"model.decoder.layers.3.router*",
	"model.decoder.layers.3.self_attn*",
	"model.decoder.layers.4.experts",
	"model.decoder.layers.4.input_layernorm",
	"model.decoder.layers.4.mlp*",
	"model.decoder.layers.4.post_attention_layernorm",
	"model.decoder.layers.4.post_feedforward_layernorm",
	"model.decoder.layers.4.post_feedforward_layernorm_1",
	"model.decoder.layers.4.post_feedforward_layernorm_2",
	"model.decoder.layers.4.pre_feedforward_layernorm",
	"model.decoder.layers.4.pre_feedforward_layernorm_2",
	"model.decoder.layers.4.router*",
	"model.decoder.layers.4.self_attn*",
	"model.decoder.layers.5.experts",
	"model.decoder.layers.5.input_layernorm",
	"model.decoder.layers.5.mlp*",
	"model.decoder.layers.5.post_attention_layernorm",
	"model.decoder.layers.5.post_feedforward_layernorm",
	"model.decoder.layers.5.post_feedforward_layernorm_1",
	"model.decoder.layers.5.post_feedforward_layernorm_2",
	"model.decoder.layers.5.pre_feedforward_layernorm",
	"model.decoder.layers.5.pre_feedforward_layernorm_2",
	"model.decoder.layers.5.router*",
	"model.decoder.layers.5.self_attn*",
	"model.decoder.layers.6.experts",
	"model.decoder.layers.6.input_layernorm",
	"model.decoder.layers.6.mlp*",
	"model.decoder.layers.6.post_attention_layernorm",
	"model.decoder.layers.6.post_feedforward_layernorm",
	"model.decoder.layers.6.post_feedforward_layernorm_1",
	"model.decoder.layers.6.post_feedforward_layernorm_2",
	"model.decoder.layers.6.pre_feedforward_layernorm",
	"model.decoder.layers.6.pre_feedforward_layernorm_2",
	"model.decoder.layers.6.router*",
	"model.decoder.layers.6.self_attn*",
	"model.decoder.layers.7.experts",
	"model.decoder.layers.7.input_layernorm",
	"model.decoder.layers.7.mlp*",
	"model.decoder.layers.7.post_attention_layernorm",
	"model.decoder.layers.7.post_feedforward_layernorm",
	"model.decoder.layers.7.post_feedforward_layernorm_1",
	"model.decoder.layers.7.post_feedforward_layernorm_2",
	"model.decoder.layers.7.pre_feedforward_layernorm",
	"model.decoder.layers.7.pre_feedforward_layernorm_2",
	"model.decoder.layers.7.router*",
	"model.decoder.layers.7.self_attn*",
	"model.decoder.layers.8.experts",
	"model.decoder.layers.8.input_layernorm",
	"model.decoder.layers.8.mlp*",
	"model.decoder.layers.8.post_attention_layernorm",
	"model.decoder.layers.8.post_feedforward_layernorm",
	"model.decoder.layers.8.post_feedforward_layernorm_1",
	"model.decoder.layers.8.post_feedforward_layernorm_2",
	"model.decoder.layers.8.pre_feedforward_layernorm",
	"model.decoder.layers.8.pre_feedforward_layernorm_2",
	"model.decoder.layers.8.router*",
	"model.decoder.layers.8.self_attn*",
	"model.decoder.layers.9.experts",
	"model.decoder.layers.9.input_layernorm",
	"model.decoder.layers.9.mlp*",
	"model.decoder.layers.9.post_attention_layernorm",
	"model.decoder.layers.9.post_feedforward_layernorm",
	"model.decoder.layers.9.post_feedforward_layernorm_1",
	"model.decoder.layers.9.post_feedforward_layernorm_2",
	"model.decoder.layers.9.pre_feedforward_layernorm",
	"model.decoder.layers.9.pre_feedforward_layernorm_2",
	"model.decoder.layers.9.router*",
	"model.decoder.layers.9.self_attn*",
	"model.decoder.norm",
	"model.decoder.self_conditioning*",
	"model.encoder.embed_vision*",
	"model.encoder.language_model.layers.0.mlp*",
	"model.encoder.language_model.layers.0.router*",
	"model.encoder.language_model.layers.0.self_attn*",
	"model.encoder.language_model.layers.1.mlp*",
	"model.encoder.language_model.layers.1.router*",
	"model.encoder.language_model.layers.1.self_attn*",
	"model.encoder.language_model.layers.10.mlp*",
	"model.encoder.language_model.layers.10.router*",
	"model.encoder.language_model.layers.10.self_attn*",
	"model.encoder.language_model.layers.11.mlp*",
	"model.encoder.language_model.layers.11.router*",
	"model.encoder.language_model.layers.11.self_attn*",
	"model.encoder.language_model.layers.12.mlp*",
	"model.encoder.language_model.layers.12.router*",
	"model.encoder.language_model.layers.12.self_attn*",
	"model.encoder.language_model.layers.13.mlp*",
	"model.encoder.language_model.layers.13.router*",
	"model.encoder.language_model.layers.13.self_attn*",
	"model.encoder.language_model.layers.14.mlp*",
	"model.encoder.language_model.layers.14.router*",
	"model.encoder.language_model.layers.14.self_attn*",
	"model.encoder.language_model.layers.15.mlp*",
	"model.encoder.language_model.layers.15.router*",
	"model.encoder.language_model.layers.15.self_attn*",
	"model.encoder.language_model.layers.16.mlp*",
	"model.encoder.language_model.layers.16.router*",
	"model.encoder.language_model.layers.16.self_attn*",
	"model.encoder.language_model.layers.17.mlp*",
	"model.encoder.language_model.layers.17.router*",
	"model.encoder.language_model.layers.17.self_attn*",
	"model.encoder.language_model.layers.18.mlp*",
	"model.encoder.language_model.layers.18.router*",
	"model.encoder.language_model.layers.18.self_attn*",
	"model.encoder.language_model.layers.19.mlp*",
	"model.encoder.language_model.layers.19.router*",
	"model.encoder.language_model.layers.19.self_attn*",
	"model.encoder.language_model.layers.2.mlp*",
	"model.encoder.language_model.layers.2.router*",
	"model.encoder.language_model.layers.2.self_attn*",
	"model.encoder.language_model.layers.20.mlp*",
	"model.encoder.language_model.layers.20.router*",
	"model.encoder.language_model.layers.20.self_attn*",
	"model.encoder.language_model.layers.21.mlp*",
	"model.encoder.language_model.layers.21.router*",
	"model.encoder.language_model.layers.21.self_attn*",
	"model.encoder.language_model.layers.22.mlp*",
	"model.encoder.language_model.layers.22.router*",
	"model.encoder.language_model.layers.22.self_attn*",
	"model.encoder.language_model.layers.23.mlp*",
	"model.encoder.language_model.layers.23.router*",
	"model.encoder.language_model.layers.23.self_attn*",
	"model.encoder.language_model.layers.24.mlp*",
	"model.encoder.language_model.layers.24.router*",
	"model.encoder.language_model.layers.24.self_attn*",
	"model.encoder.language_model.layers.25.mlp*",
	"model.encoder.language_model.layers.25.router*",
	"model.encoder.language_model.layers.25.self_attn*",
	"model.encoder.language_model.layers.26.mlp*",
	"model.encoder.language_model.layers.26.router*",
	"model.encoder.language_model.layers.26.self_attn*",
	"model.encoder.language_model.layers.27.mlp*",
	"model.encoder.language_model.layers.27.router*",
	"model.encoder.language_model.layers.27.self_attn*",
	"model.encoder.language_model.layers.28.mlp*",
	"model.encoder.language_model.layers.28.router*",
	"model.encoder.language_model.layers.28.self_attn*",
	"model.encoder.language_model.layers.29.mlp*",
	"model.encoder.language_model.layers.29.router*",
	"model.encoder.language_model.layers.29.self_attn*",
	"model.encoder.language_model.layers.3.mlp*",
	"model.encoder.language_model.layers.3.router*",
	"model.encoder.language_model.layers.3.self_attn*",
	"model.encoder.language_model.layers.4.mlp*",
	"model.encoder.language_model.layers.4.router*",
	"model.encoder.language_model.layers.4.self_attn*",
	"model.encoder.language_model.layers.5.mlp*",
	"model.encoder.language_model.layers.5.router*",
	"model.encoder.language_model.layers.5.self_attn*",
	"model.encoder.language_model.layers.6.mlp*",
	"model.encoder.language_model.layers.6.router*",
	"model.encoder.language_model.layers.6.self_attn*",
	"model.encoder.language_model.layers.7.mlp*",
	"model.encoder.language_model.layers.7.router*",
	"model.encoder.language_model.layers.7.self_attn*",
	"model.encoder.language_model.layers.8.mlp*",
	"model.encoder.language_model.layers.8.router*",
	"model.encoder.language_model.layers.8.self_attn*",
	"model.encoder.language_model.layers.9.mlp*",
	"model.encoder.language_model.layers.9.router*",
	"model.encoder.language_model.layers.9.self_attn*",
	"model.encoder.vision_tower*"
	]
	}
	}