Qwen2.5-7B-AutoRound-W4A16-RTN

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of Qwen/Qwen2.5-7B generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	Qwen/Qwen2.5-7B
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	5313 MB

Evaluation Results

Task	Accuracy
hellaswag	0.5862
mmlu	0.6995
mmlu_abstract_algebra	0.5000
mmlu_anatomy	0.6741
mmlu_astronomy	0.8092
mmlu_business_ethics	0.7700
mmlu_clinical_knowledge	0.7698
mmlu_college_biology	0.8194
mmlu_college_chemistry	0.5300
mmlu_college_computer_science	0.6200
mmlu_college_mathematics	0.5200
mmlu_college_medicine	0.6936
mmlu_college_physics	0.5098
mmlu_computer_security	0.8500
mmlu_conceptual_physics	0.7447
mmlu_econometrics	0.5965
mmlu_electrical_engineering	0.6966
mmlu_elementary_mathematics	0.6878
mmlu_formal_logic	0.5317
mmlu_global_facts	0.4100
mmlu_high_school_biology	0.8548
mmlu_high_school_chemistry	0.6305
mmlu_high_school_computer_science	0.8300
mmlu_high_school_european_history	0.8000
mmlu_high_school_geography	0.8737
mmlu_high_school_government_and_politics	0.9275
mmlu_high_school_macroeconomics	0.7821
mmlu_high_school_mathematics	0.5259
mmlu_high_school_microeconomics	0.8655
mmlu_high_school_physics	0.5695
mmlu_high_school_psychology	0.8807
mmlu_high_school_statistics	0.6620
mmlu_high_school_us_history	0.8725
mmlu_high_school_world_history	0.8143
mmlu_human_aging	0.7578
mmlu_human_sexuality	0.8244
mmlu_humanities	0.5977
mmlu_international_law	0.8017
mmlu_jurisprudence	0.8241
mmlu_logical_fallacies	0.8160
mmlu_machine_learning	0.6429
mmlu_management	0.9029
mmlu_marketing	0.9402
mmlu_medical_genetics	0.8300
mmlu_miscellaneous	0.8544
mmlu_moral_disputes	0.7861
mmlu_moral_scenarios	0.2525
mmlu_nutrition	0.8039
mmlu_other	0.7612
mmlu_philosophy	0.7556
mmlu_prehistory	0.8117
mmlu_professional_accounting	0.5390
mmlu_professional_law	0.5098
mmlu_professional_medicine	0.7279
mmlu_professional_psychology	0.7533
mmlu_public_relations	0.7364
mmlu_security_studies	0.7633
mmlu_social_sciences	0.8141
mmlu_sociology	0.8458
mmlu_stem	0.6790
mmlu_us_foreign_policy	0.8700
mmlu_virology	0.5602
mmlu_world_religions	0.8480
piqa	0.7835

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 = "Qwen2.5-7B-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 Qwen2.5-7B-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:

• Intel Neural Compressor
• AutoRound

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.