Qwen3.5-4B-AutoRound-MXFP8-Tuning

Model Details

This model is a MXFP8 quantization of Qwen/Qwen3.5-4B generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	Qwen/Qwen3.5-4B
Quantization Tool	TUNING
Quantization Scheme	MXFP8
Quantized Size	5591 MB

Evaluation Results

Task	Accuracy
hellaswag	0.5410
mmlu	0.7379
mmlu_abstract_algebra	0.5800
mmlu_anatomy	0.6963
mmlu_astronomy	0.8553
mmlu_business_ethics	0.7600
mmlu_clinical_knowledge	0.8038
mmlu_college_biology	0.8681
mmlu_college_chemistry	0.5700
mmlu_college_computer_science	0.7700
mmlu_college_mathematics	0.5900
mmlu_college_medicine	0.7630
mmlu_college_physics	0.6078
mmlu_computer_security	0.8500
mmlu_conceptual_physics	0.8511
mmlu_econometrics	0.6667
mmlu_electrical_engineering	0.8069
mmlu_elementary_mathematics	0.7328
mmlu_formal_logic	0.6270
mmlu_global_facts	0.4600
mmlu_high_school_biology	0.9194
mmlu_high_school_chemistry	0.7635
mmlu_high_school_computer_science	0.8100
mmlu_high_school_european_history	0.8303
mmlu_high_school_geography	0.8636
mmlu_high_school_government_and_politics	0.9378
mmlu_high_school_macroeconomics	0.7846
mmlu_high_school_mathematics	0.4889
mmlu_high_school_microeconomics	0.9118
mmlu_high_school_physics	0.6954
mmlu_high_school_psychology	0.9174
mmlu_high_school_statistics	0.7130
mmlu_high_school_us_history	0.8529
mmlu_high_school_world_history	0.8439
mmlu_human_aging	0.7130
mmlu_human_sexuality	0.8702
mmlu_humanities	0.6540
mmlu_international_law	0.8678
mmlu_jurisprudence	0.8333
mmlu_logical_fallacies	0.8221
mmlu_machine_learning	0.5714
mmlu_management	0.8835
mmlu_marketing	0.9103
mmlu_medical_genetics	0.8600
mmlu_miscellaneous	0.8416
mmlu_moral_disputes	0.7486
mmlu_moral_scenarios	0.4760
mmlu_nutrition	0.8105
mmlu_other	0.7763
mmlu_philosophy	0.7588
mmlu_prehistory	0.8241
mmlu_professional_accounting	0.6099
mmlu_professional_law	0.5417
mmlu_professional_medicine	0.8162
mmlu_professional_psychology	0.7614
mmlu_public_relations	0.7182
mmlu_security_studies	0.7510
mmlu_social_sciences	0.8304
mmlu_sociology	0.8856
mmlu_stem	0.7349
mmlu_us_foreign_policy	0.8300
mmlu_virology	0.5723
mmlu_world_religions	0.8129
piqa	0.7715

How to Use

HF Usage

Step 1: Install AutoRound

pip install auto-round

Step 2: Load and run the quantized model

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Qwen3.5-4B-AutoRound-MXFP8-Tuning"

# load the tokenizer and the model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")

# prepare the model input
prompt = "Write a quick sort algorithm."
messages = [{"role": "user", "content": prompt}]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# conduct text completion
generated_ids = model.generate(**model_inputs, max_new_tokens=512)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]) :].tolist()

content = tokenizer.decode(output_ids, skip_special_tokens=True)
print("content:", content)

VLLM Usage

vllm serve Qwen3.5-4B-AutoRound-MXFP8-Tuning \
    --trust-remote-code \
    --dtype bfloat16 \
    --tensor_parallel_size 1

If you encounter any issues, feel free to open an issue on the AutoRound GitHub repo or provide feedback on the Low-Bit Open LLM Leaderboard.

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

This model is part of the Intel Low-Bit Open LLM Leaderboard initiative.