gpt-oss-20b-AutoRound-MXFP4-RTN

Model Details

This model is a MXFP4 (Microscaling FP4) quantization of openai/gpt-oss-20b generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	openai/gpt-oss-20b
Quantization Tool	AutoRound
Quantization Scheme	MXFP4
Quantized Size	13125 MB

Evaluation Results

Task	Accuracy
hellaswag	0.2581
mmlu	0.2488
mmlu_abstract_algebra	0.3100
mmlu_anatomy	0.2889
mmlu_astronomy	0.2434
mmlu_business_ethics	0.2400
mmlu_clinical_knowledge	0.2340
mmlu_college_biology	0.2500
mmlu_college_chemistry	0.2000
mmlu_college_computer_science	0.2800
mmlu_college_mathematics	0.3100
mmlu_college_medicine	0.1908
mmlu_college_physics	0.1863
mmlu_computer_security	0.2400
mmlu_conceptual_physics	0.2426
mmlu_econometrics	0.2368
mmlu_electrical_engineering	0.2207
mmlu_elementary_mathematics	0.2222
mmlu_formal_logic	0.2937
mmlu_global_facts	0.2400
mmlu_high_school_biology	0.2032
mmlu_high_school_chemistry	0.2414
mmlu_high_school_computer_science	0.2700
mmlu_high_school_european_history	0.2848
mmlu_high_school_geography	0.2323
mmlu_high_school_government_and_politics	0.2280
mmlu_high_school_macroeconomics	0.2487
mmlu_high_school_mathematics	0.2593
mmlu_high_school_microeconomics	0.2311
mmlu_high_school_physics	0.2781
mmlu_high_school_psychology	0.2220
mmlu_high_school_statistics	0.2361
mmlu_high_school_us_history	0.2304
mmlu_high_school_world_history	0.2236
mmlu_human_aging	0.2915
mmlu_human_sexuality	0.2443
mmlu_humanities	0.2563
mmlu_international_law	0.3058
mmlu_jurisprudence	0.2593
mmlu_logical_fallacies	0.3067
mmlu_machine_learning	0.2589
mmlu_management	0.2136
mmlu_marketing	0.2863
mmlu_medical_genetics	0.2600
mmlu_miscellaneous	0.2861
mmlu_moral_disputes	0.2312
mmlu_moral_scenarios	0.2547
mmlu_nutrition	0.2222
mmlu_other	0.2514
mmlu_philosophy	0.2572
mmlu_prehistory	0.2377
mmlu_professional_accounting	0.2376
mmlu_professional_law	0.2634
mmlu_professional_medicine	0.1985
mmlu_professional_psychology	0.2402
mmlu_public_relations	0.3364
mmlu_security_studies	0.2367
mmlu_social_sciences	0.2398
mmlu_sociology	0.2537
mmlu_stem	0.2439
mmlu_us_foreign_policy	0.2300
mmlu_virology	0.2711
mmlu_world_religions	0.2222
piqa	0.5245

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 = "gpt-oss-20b-AutoRound-MXFP4-RTN"

# 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 gpt-oss-20b-AutoRound-MXFP4-RTN \
    --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:

• Intel Neural Compressor
• AutoRound

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.