Instructions to use intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound with Transformers:

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

pipe = pipeline("text-generation", model="intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound")
model = AutoModelForCausalLM.from_pretrained("intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound

SGLang

How to use intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound 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 "intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound with Docker Model Runner:
```
docker model run hf.co/intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound
```

Model Details

This model is an MXFP4 (mixed MXFP4 experts / MXFP8 linear) model of deepseek-ai/DeepSeek-V4-Flash-0731 generated by intel/auto-round using RTN. Please follow the license of the original model.

Quantized (MXFP4): routed experts
Kept in BF16: router / embeddings / lm_head

How to Use

Environment

pip install vllm==0.29.0
uv pip install git+https://github.com/vllm-project/vllm.git@1b6e277ac010fb3361369162c42d39b06d612e24

vLLM Usage

vllm serve intel-ai/DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound

Evaluation

Benchmark	BF16	MXFP4	Recovery
GSM8K	95.83	95.68	99.84%
MMLU	86.49	86.31	99.79%
PIQA	83.41	82.32	98.69%
HellaSwag	68.18	67.88	99.56%
Average	83.48	83.05	99.48%

Generate the Model

The original AutoRound version and source revision were not recorded for this checkpoint. The DeepSeek-V4 Flash/Pro generation PR is not evidence of this checkpoint's toolchain; obtain its exact build from the model maintainer for end-to-end regeneration.

auto-round \
    --model_name deepseek-ai/DeepSeek-V4-Flash-0731 \
    --model_free \
    --scheme MXFP8 \
    --ignore_layers "compressor,indexer.weights_proj" \
    --layer_config "{ffn.experts:{bits:4,data_type:mx_fp}}" \
    --format llm_compressor \
    --output_dir DeepSeek-V4-Flash-0731-MXFP4-Mixed-CT-AutoRound

Ethical Considerations and Limitations

The model can produce factually incorrect output, and should not be relied on to produce factually accurate information. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs. Therefore, before deploying any applications of the model, developers should perform safety testing.

Caveats and Recommendations

The vLLM PR pinned above supports a BF16 fallback for the compressed-tensors output projection. The reported evaluation instead used an MXFP8 GEMM path whose exact build was not recorded; this installation must not be treated as that original evaluation environment.

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Here are a couple of useful links to learn more about Intel's AI software:

Disclaimer

The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please consult an attorney before using this model for commercial purposes.

Cite

@article{cheng2023optimize,
  title={Optimize weight rounding via signed gradient descent for the quantization of llms},
  author={Cheng, Wenhua and Zhang, Weiwei and Shen, Haihao and Cai, Yiyang and He, Xin and Lv, Kaokao and Liu, Yi},
  journal={arXiv preprint arXiv:2309.05516},
  year={2023}
}

arxiv github

Signed-off-by: Lv, Kaokao kaokao.lv@intel.com