gemma-4-E2B-it-qat-q4_0-unquantized-AutoRound-W4A16-RTN

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of google/gemma-4-E2B-it-qat-q4_0-unquantized generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	google/gemma-4-E2B-it-qat-q4_0-unquantized
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	7106 MB

Evaluation Results

Task	Accuracy
hellaswag	0.2985
mmlu	0.2940
mmlu_abstract_algebra	0.2000
mmlu_anatomy	0.2444
mmlu_astronomy	0.3750
mmlu_business_ethics	0.2700
mmlu_clinical_knowledge	0.3170
mmlu_college_biology	0.3125
mmlu_college_chemistry	0.3500
mmlu_college_computer_science	0.3000
mmlu_college_mathematics	0.2300
mmlu_college_medicine	0.3584
mmlu_college_physics	0.2451
mmlu_computer_security	0.1600
mmlu_conceptual_physics	0.2638
mmlu_econometrics	0.2632
mmlu_electrical_engineering	0.2897
mmlu_elementary_mathematics	0.2857
mmlu_formal_logic	0.2698
mmlu_global_facts	0.1800
mmlu_high_school_biology	0.2968
mmlu_high_school_chemistry	0.2660
mmlu_high_school_computer_science	0.2200
mmlu_high_school_european_history	0.4424
mmlu_high_school_geography	0.2626
mmlu_high_school_government_and_politics	0.3109
mmlu_high_school_macroeconomics	0.3231
mmlu_high_school_mathematics	0.2556
mmlu_high_school_microeconomics	0.3277
mmlu_high_school_physics	0.3046
mmlu_high_school_psychology	0.3229
mmlu_high_school_statistics	0.2500
mmlu_high_school_us_history	0.4167
mmlu_high_school_world_history	0.5232
mmlu_human_aging	0.2287
mmlu_human_sexuality	0.2748
mmlu_humanities	0.3048
mmlu_international_law	0.3140
mmlu_jurisprudence	0.3704
mmlu_logical_fallacies	0.2393
mmlu_machine_learning	0.2321
mmlu_management	0.3786
mmlu_marketing	0.2863
mmlu_medical_genetics	0.3400
mmlu_miscellaneous	0.2261
mmlu_moral_disputes	0.3092
mmlu_moral_scenarios	0.2760
mmlu_nutrition	0.3366
mmlu_other	0.2836
mmlu_philosophy	0.2958
mmlu_prehistory	0.2778
mmlu_professional_accounting	0.2624
mmlu_professional_law	0.2731
mmlu_professional_medicine	0.3382
mmlu_professional_psychology	0.2729
mmlu_public_relations	0.3091
mmlu_security_studies	0.3918
mmlu_social_sciences	0.3100
mmlu_sociology	0.3333
mmlu_stem	0.2724
mmlu_us_foreign_policy	0.3200
mmlu_virology	0.3193
mmlu_world_religions	0.2690
piqa	0.5637

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-E2B-it-qat-q4_0-unquantized-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-E2B-it-qat-q4_0-unquantized-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.