gemma-4-E4B-it-AutoRound-W4A16-RTN

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of google/gemma-4-E4B-it generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	google/gemma-4-E4B-it
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	9576 MB

Evaluation Results

Task	Accuracy
hellaswag	0.2991
mmlu	0.3842
mmlu_abstract_algebra	0.3000
mmlu_anatomy	0.3185
mmlu_astronomy	0.3750
mmlu_business_ethics	0.4400
mmlu_clinical_knowledge	0.3736
mmlu_college_biology	0.4306
mmlu_college_chemistry	0.2800
mmlu_college_computer_science	0.3500
mmlu_college_mathematics	0.3200
mmlu_college_medicine	0.3873
mmlu_college_physics	0.2843
mmlu_computer_security	0.4200
mmlu_conceptual_physics	0.3404
mmlu_econometrics	0.3246
mmlu_electrical_engineering	0.3586
mmlu_elementary_mathematics	0.3545
mmlu_formal_logic	0.3730
mmlu_global_facts	0.2400
mmlu_high_school_biology	0.4258
mmlu_high_school_chemistry	0.3202
mmlu_high_school_computer_science	0.4100
mmlu_high_school_european_history	0.5939
mmlu_high_school_geography	0.3434
mmlu_high_school_government_and_politics	0.4508
mmlu_high_school_macroeconomics	0.3821
mmlu_high_school_mathematics	0.2667
mmlu_high_school_microeconomics	0.3824
mmlu_high_school_physics	0.3444
mmlu_high_school_psychology	0.4018
mmlu_high_school_statistics	0.3935
mmlu_high_school_us_history	0.5343
mmlu_high_school_world_history	0.5359
mmlu_human_aging	0.3856
mmlu_human_sexuality	0.4504
mmlu_humanities	0.3883
mmlu_international_law	0.6033
mmlu_jurisprudence	0.4167
mmlu_logical_fallacies	0.3681
mmlu_machine_learning	0.4375
mmlu_management	0.3689
mmlu_marketing	0.4829
mmlu_medical_genetics	0.4700
mmlu_miscellaneous	0.3972
mmlu_moral_disputes	0.4191
mmlu_moral_scenarios	0.2615
mmlu_nutrition	0.4085
mmlu_other	0.3888
mmlu_philosophy	0.3698
mmlu_prehistory	0.3519
mmlu_professional_accounting	0.3511
mmlu_professional_law	0.3774
mmlu_professional_medicine	0.3566
mmlu_professional_psychology	0.3987
mmlu_public_relations	0.3182
mmlu_security_studies	0.4163
mmlu_social_sciences	0.4030
mmlu_sociology	0.4826
mmlu_stem	0.3552
mmlu_us_foreign_policy	0.5200
mmlu_virology	0.3494
mmlu_world_religions	0.4737
piqa	0.5740

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-E4B-it-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 gemma-4-E4B-it-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.