Instructions to use sojufx/Qwopus3.8-27B-Flash-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sojufx/Qwopus3.8-27B-Flash-NVFP4 with Transformers:

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

pipe = pipeline("image-text-to-text", model="sojufx/Qwopus3.8-27B-Flash-NVFP4")
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("sojufx/Qwopus3.8-27B-Flash-NVFP4")
model = AutoModelForMultimodalLM.from_pretrained("sojufx/Qwopus3.8-27B-Flash-NVFP4", 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 sojufx/Qwopus3.8-27B-Flash-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sojufx/Qwopus3.8-27B-Flash-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sojufx/Qwopus3.8-27B-Flash-NVFP4",
		"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/sojufx/Qwopus3.8-27B-Flash-NVFP4

SGLang

How to use sojufx/Qwopus3.8-27B-Flash-NVFP4 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 "sojufx/Qwopus3.8-27B-Flash-NVFP4" \
    --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": "sojufx/Qwopus3.8-27B-Flash-NVFP4",
		"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 "sojufx/Qwopus3.8-27B-Flash-NVFP4" \
        --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": "sojufx/Qwopus3.8-27B-Flash-NVFP4",
		"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 sojufx/Qwopus3.8-27B-Flash-NVFP4 with Docker Model Runner:
```
docker model run hf.co/sojufx/Qwopus3.8-27B-Flash-NVFP4
```

Qwopus3.8-27B-Flash-NVFP4 / hf_quant_config.json

sojufx

Upload ModelOpt NVFP4 conversion

892f62b verified 2 months ago

Raw History Blame Contribute Delete

54.8 kB

{
    "producer": {
        "name": "modelopt",
        "version": "0.37.0"
    },
    "quantization": {
        "quant_algo": "MIXED_PRECISION",
        "kv_cache_quant_algo": "FP8",
        "quantized_layers": {
            "model.language_model.layers.0.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.0.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.0.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.0.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.0.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.0.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.1.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.1.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.1.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.1.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.1.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.1.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.2.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.2.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.2.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.2.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.2.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.2.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.3.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.3.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.3.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.4.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.4.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.4.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.4.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.4.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.4.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.5.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.5.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.5.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.5.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.5.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.5.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.6.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.6.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.6.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.6.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.6.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.6.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.7.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.7.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.7.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.7.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.7.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.7.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.7.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.8.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.8.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.8.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.8.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.8.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.8.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.9.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.9.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.9.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.9.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.9.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.9.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.10.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.10.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.10.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.10.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.10.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.10.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.11.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.11.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.11.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.11.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.11.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.11.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.11.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.12.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.12.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.12.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.12.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.12.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.12.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.13.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.13.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.13.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.13.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.13.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.13.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.14.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.14.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.14.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.14.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.14.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.14.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.15.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.15.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.15.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.15.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.15.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.15.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.15.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.16.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.16.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.16.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.16.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.16.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.16.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.17.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.17.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.17.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.17.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.17.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.17.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.18.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.18.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.18.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.18.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.18.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.18.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.19.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.19.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.19.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.19.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.19.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.19.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.19.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.20.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.20.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.20.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.20.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.20.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.20.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.21.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.21.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.21.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.21.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.21.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.21.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.22.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.22.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.22.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.22.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.22.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.22.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.23.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.23.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.23.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.23.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.23.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.23.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.23.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.24.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.24.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.24.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.24.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.24.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.24.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.25.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.25.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.25.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.25.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.25.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.25.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.26.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.26.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.26.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.26.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.26.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.26.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.27.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.27.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.27.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.27.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.27.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.27.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.27.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.28.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.28.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.28.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.28.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.28.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.28.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.29.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.29.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.29.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.29.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.29.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.29.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.30.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.30.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.30.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.30.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.30.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.30.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.31.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.31.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.31.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.31.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.31.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.31.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.31.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.32.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.32.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.32.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.32.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.32.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.32.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.33.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.33.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.33.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.33.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.33.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.33.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.34.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.34.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.34.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.34.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.34.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.34.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.35.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.35.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.35.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.35.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.35.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.35.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.35.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.36.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.36.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.36.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.36.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.36.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.36.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.37.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.37.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.37.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.37.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.37.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.37.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.38.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.38.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.38.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.38.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.38.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.38.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.39.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.39.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.39.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.39.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.39.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.39.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.39.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.40.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.40.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.40.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.40.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.40.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.40.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.41.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.41.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.41.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.41.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.41.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.41.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.42.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.42.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.42.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.42.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.42.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.42.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.43.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.43.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.43.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.43.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.43.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.43.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.43.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.44.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.44.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.44.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.44.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.44.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.44.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.45.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.45.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.45.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.45.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.45.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.45.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.46.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.46.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.46.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.46.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.46.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.46.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.47.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.47.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.47.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.47.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.47.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.47.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.47.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.48.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.48.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.48.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.48.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.48.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.48.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.49.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.49.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.49.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.49.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.49.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.49.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.50.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.50.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.50.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.50.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.50.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.50.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.51.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.51.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.51.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.51.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.51.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.51.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.51.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.52.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.52.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.52.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.52.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.52.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.52.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.53.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.53.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.53.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.53.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.53.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.53.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.54.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.54.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.54.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.54.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.54.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.54.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.55.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.55.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.55.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.55.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.55.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.55.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.55.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.56.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.56.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.56.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.56.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.56.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.56.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.57.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.57.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.57.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.57.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.57.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.57.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.58.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.58.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.58.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.58.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.58.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.58.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.59.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.59.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.59.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.59.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.59.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.59.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.59.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.60.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.60.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.60.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.60.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.60.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.60.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.61.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.61.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.61.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.61.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.61.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.61.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.62.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.62.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.62.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.62.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.62.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.62.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.63.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.63.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.63.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.63.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.63.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.63.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.63.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            }
        },
        "exclude_modules": [
            "mtp*",
            "mtp.layers.0*"
        ]
    }
}
{
    "producer": {
        "name": "modelopt",
        "version": "0.37.0"
    },
    "quantization": {
        "quant_algo": "MIXED_PRECISION",
        "kv_cache_quant_algo": "FP8",
        "quantized_layers": {
            "model.language_model.layers.0.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.0.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.0.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.0.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.0.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.0.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.1.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.1.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.1.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.1.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.1.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.1.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.2.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.2.linear_attn.in_proj_qkv": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.2.linear_attn.in_proj_z": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.2.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.2.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.2.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.3.self_attn.q_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.self_attn.k_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.self_attn.v_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.self_attn.o_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.3.mlp.gate_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.3.mlp.up_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.3.mlp.down_proj": {
                "quant_algo": "W4A16_NVFP4",
                "group_size": 16
            },
            "model.language_model.layers.4.linear_attn.out_proj": {
                "quant_algo": "FP8"
            },
            "model.language_model.layers.4.linear_attn.in_proj_qkv": {