Gemma4-GarnetV2-31B-AutoRound-W4A16-RTN

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of ConicCat/Gemma4-GarnetV2-31B generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	ConicCat/Gemma4-GarnetV2-31B
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	18299 MB

Evaluation Results

Task	Accuracy
hellaswag	0.3168
mmlu	0.4471
mmlu_abstract_algebra	0.3600
mmlu_anatomy	0.4296
mmlu_astronomy	0.4408
mmlu_business_ethics	0.3700
mmlu_clinical_knowledge	0.4981
mmlu_college_biology	0.5139
mmlu_college_chemistry	0.4100
mmlu_college_computer_science	0.4000
mmlu_college_mathematics	0.3400
mmlu_college_medicine	0.3931
mmlu_college_physics	0.4020
mmlu_computer_security	0.4800
mmlu_conceptual_physics	0.3574
mmlu_econometrics	0.2895
mmlu_electrical_engineering	0.3586
mmlu_elementary_mathematics	0.4127
mmlu_formal_logic	0.4603
mmlu_global_facts	0.2700
mmlu_high_school_biology	0.5710
mmlu_high_school_chemistry	0.3941
mmlu_high_school_computer_science	0.4900
mmlu_high_school_european_history	0.7091
mmlu_high_school_geography	0.5808
mmlu_high_school_government_and_politics	0.5803
mmlu_high_school_macroeconomics	0.4897
mmlu_high_school_mathematics	0.2852
mmlu_high_school_microeconomics	0.4958
mmlu_high_school_physics	0.4570
mmlu_high_school_psychology	0.5890
mmlu_high_school_statistics	0.6157
mmlu_high_school_us_history	0.7010
mmlu_high_school_world_history	0.7426
mmlu_human_aging	0.3453
mmlu_human_sexuality	0.4656
mmlu_humanities	0.4372
mmlu_international_law	0.3223
mmlu_jurisprudence	0.3611
mmlu_logical_fallacies	0.4479
mmlu_machine_learning	0.2232
mmlu_management	0.5243
mmlu_marketing	0.5427
mmlu_medical_genetics	0.4800
mmlu_miscellaneous	0.4036
mmlu_moral_disputes	0.3179
mmlu_moral_scenarios	0.3385
mmlu_nutrition	0.4608
mmlu_other	0.4377
mmlu_philosophy	0.4244
mmlu_prehistory	0.4506
mmlu_professional_accounting	0.3652
mmlu_professional_law	0.4185
mmlu_professional_medicine	0.6434
mmlu_professional_psychology	0.4150
mmlu_public_relations	0.4182
mmlu_security_studies	0.4857
mmlu_social_sciences	0.4940
mmlu_sociology	0.5423
mmlu_stem	0.4253
mmlu_us_foreign_policy	0.4100
mmlu_virology	0.3313
mmlu_world_religions	0.4620
piqa	0.5827

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 = "Gemma4-GarnetV2-31B-AutoRound-W4A16-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 Gemma4-GarnetV2-31B-AutoRound-W4A16-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.