gemma-4-12B-AutoRound-MXFP4-RTN

Model Details

This model is a MXFP4 (Microscaling FP4) quantization of google/gemma-4-12B generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	google/gemma-4-12B
Quantization Tool	AutoRound
Quantization Scheme	MXFP4
Quantized Size	10915 MB

Evaluation Results

Task	Accuracy
hellaswag	0.5902
mmlu	0.6576
mmlu_abstract_algebra	0.3600
mmlu_anatomy	0.6519
mmlu_astronomy	0.7763
mmlu_business_ethics	0.6600
mmlu_clinical_knowledge	0.7208
mmlu_college_biology	0.7569
mmlu_college_chemistry	0.4900
mmlu_college_computer_science	0.5700
mmlu_college_mathematics	0.3200
mmlu_college_medicine	0.7110
mmlu_college_physics	0.4510
mmlu_computer_security	0.7200
mmlu_conceptual_physics	0.6170
mmlu_econometrics	0.4386
mmlu_electrical_engineering	0.6828
mmlu_elementary_mathematics	0.5053
mmlu_formal_logic	0.3968
mmlu_global_facts	0.4000
mmlu_high_school_biology	0.8323
mmlu_high_school_chemistry	0.6059
mmlu_high_school_computer_science	0.6900
mmlu_high_school_european_history	0.7697
mmlu_high_school_geography	0.8687
mmlu_high_school_government_and_politics	0.8601
mmlu_high_school_macroeconomics	0.7103
mmlu_high_school_mathematics	0.3815
mmlu_high_school_microeconomics	0.7605
mmlu_high_school_physics	0.4570
mmlu_high_school_psychology	0.8385
mmlu_high_school_statistics	0.6250
mmlu_high_school_us_history	0.8284
mmlu_high_school_world_history	0.8565
mmlu_human_aging	0.7040
mmlu_human_sexuality	0.8244
mmlu_humanities	0.5872
mmlu_international_law	0.8595
mmlu_jurisprudence	0.7593
mmlu_logical_fallacies	0.7853
mmlu_machine_learning	0.4643
mmlu_management	0.8058
mmlu_marketing	0.9017
mmlu_medical_genetics	0.7000
mmlu_miscellaneous	0.8212
mmlu_moral_disputes	0.7601
mmlu_moral_scenarios	0.2771
mmlu_nutrition	0.7810
mmlu_other	0.7203
mmlu_philosophy	0.7106
mmlu_prehistory	0.7778
mmlu_professional_accounting	0.5071
mmlu_professional_law	0.5039
mmlu_professional_medicine	0.7059
mmlu_professional_psychology	0.7451
mmlu_public_relations	0.6818
mmlu_security_studies	0.7347
mmlu_social_sciences	0.7741
mmlu_sociology	0.8458
mmlu_stem	0.5871
mmlu_us_foreign_policy	0.9000
mmlu_virology	0.4819
mmlu_world_religions	0.8363
piqa	0.7916

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 = "gemma-4-12B-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 gemma-4-12B-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:

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.