Qwen3-1.7B-AutoRound-W4A16-RTN

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of Qwen/Qwen3-1.7B generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	Qwen/Qwen3-1.7B
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	1292 MB

Evaluation Results

Task	Accuracy
hellaswag	0.4451
mmlu	0.5391
mmlu_abstract_algebra	0.3600
mmlu_anatomy	0.5111
mmlu_astronomy	0.6053
mmlu_business_ethics	0.5400
mmlu_clinical_knowledge	0.5811
mmlu_college_biology	0.6528
mmlu_college_chemistry	0.4300
mmlu_college_computer_science	0.4300
mmlu_college_mathematics	0.4000
mmlu_college_medicine	0.5549
mmlu_college_physics	0.3039
mmlu_computer_security	0.7000
mmlu_conceptual_physics	0.6085
mmlu_econometrics	0.5263
mmlu_electrical_engineering	0.5448
mmlu_elementary_mathematics	0.4947
mmlu_formal_logic	0.4603
mmlu_global_facts	0.2200
mmlu_high_school_biology	0.6839
mmlu_high_school_chemistry	0.5369
mmlu_high_school_computer_science	0.6500
mmlu_high_school_european_history	0.7030
mmlu_high_school_geography	0.6717
mmlu_high_school_government_and_politics	0.6529
mmlu_high_school_macroeconomics	0.5308
mmlu_high_school_mathematics	0.3926
mmlu_high_school_microeconomics	0.6008
mmlu_high_school_physics	0.4238
mmlu_high_school_psychology	0.7468
mmlu_high_school_statistics	0.5231
mmlu_high_school_us_history	0.6275
mmlu_high_school_world_history	0.7046
mmlu_human_aging	0.6502
mmlu_human_sexuality	0.6336
mmlu_humanities	0.4678
mmlu_international_law	0.6446
mmlu_jurisprudence	0.6111
mmlu_logical_fallacies	0.6442
mmlu_machine_learning	0.4018
mmlu_management	0.6505
mmlu_marketing	0.7821
mmlu_medical_genetics	0.6200
mmlu_miscellaneous	0.6654
mmlu_moral_disputes	0.5838
mmlu_moral_scenarios	0.2425
mmlu_nutrition	0.5719
mmlu_other	0.5855
mmlu_philosophy	0.6077
mmlu_prehistory	0.5833
mmlu_professional_accounting	0.4326
mmlu_professional_law	0.3664
mmlu_professional_medicine	0.5331
mmlu_professional_psychology	0.5523
mmlu_public_relations	0.5273
mmlu_security_studies	0.6245
mmlu_social_sciences	0.6204
mmlu_sociology	0.6269
mmlu_stem	0.5205
mmlu_us_foreign_policy	0.7500
mmlu_virology	0.4398
mmlu_world_religions	0.7251
piqa	0.7067

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 = "Qwen3-1.7B-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 Qwen3-1.7B-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.