How to Run Locally
vLLM is supported here: https://github.com/vllm-project/vllm/pull/45645
CUDA_VISIBLE_DEVICES=0,1 vllm serve INCModel/DeepSeek-V4-Flash-MXFP4-Mixed-CT-AutoRound \
--trust-remote-code \
--kv-cache-dtype fp8 \
--block-size 256 \
--tensor-parallel-size 2 \
--attention_config.use_fp4_indexer_cache=True \
--port 8009 \
--no-enable-flashinfer-autotune \
--enforce-eager
curl -s http://127.0.0.1:8009/v1/chat/completions -H "Content-Type: application/json" -d '{
"model": "DeepSeek-V4-Flash-MXFP4-Mixed-CT-AutoRound",
"messages": [
{"role":"user","content":"2+3=?"}
],
"max_tokens": 10,
"extra_body": {
"chat_template_kwargs": {
"enable_thinking": true
}
}
}' | python3 -m json.tool
For local deployment, we recommend setting the sampling parameters to temperature = 1.0, top_p = 1.0. For the Think Max reasoning mode, we recommend setting the context window to at least 384K tokens.
Generate the Model
Depends on this PR: https://github.com/intel/auto-round/pull/1921
auto-round deepseek-ai/DeepSeek-V4-Flash \
--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-MXFP4-Mixed"
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
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