GLM-5.2-MXFP4-Mixed-CT-AutoRound

Model Details

This model is a MXFP4 and MXFP8 mixed model of zai-org/GLM-5.2 generated by intel/auto-round with RTN mode. The model format is Compressed Tensor (CT), fully compatible with vLLM.

<!-- Not real data | GLM-5.2 | gsm8k | mmlu | piqa | hellaswag | avg | ratio | |-------------------|---------|---------|---------|------------|--------|--------| | Raw | 0.9386 | 0.8898 | 0.8471 | 0.7747 | 0.863 | | | **INCModel2/GLM-5.2-MXFP4-Mixed-LLMC** | 0.9439 | 0.8847 | 0.8411 | 0.7646 | 0.859 | 99.54% | -->

vllm Infernece Example

vllm serve INCModel2/GLM-5.2-MXFP4-Mixed-LLMC \
  --kv-cache-dtype fp8_e4m3 \
  --tensor-parallel-size 4 \
  --tool-call-parser glm47 \
  --enable-auto-tool-choice \
  --reasoning-parser glm45 \
  --served-model-name glm \
  --max-model-len 4096

curl http://localhost:8000/v1/chat/completions -H "Content-Type: application/json" -d ' {
    "model": "glm",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Write code to fine-tune an LLM."}
    ],
    "temperature": 1,
    "max_tokens": 2048
  } '

Generate the Model

auto-round zai-org/GLM-5.2 --model_free --format llm_compressor --scheme MXFP8 --output_dir /workspace/models/zai-org/GLM-5.2-MXFP4-MXFP8  --ignore_layers layers.0,layers.1,layers.2,indexer.weights_proj,indexer.wk  --layer_config "{mlp.experts:{scheme:MXFP4}}"

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:

• Intel Neural Compressor

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