This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of Qwen/Qwen-Image-Bench generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	Qwen/Qwen-Image-Bench
Quantization Tool	TUNING
Quantization Scheme	W4A16
Quantized Size	17799 MB

Evaluation Results

Task	Accuracy
hellaswag	0.6345
mmlu	0.8260
mmlu_abstract_algebra	0.7600
mmlu_anatomy	0.8296
mmlu_astronomy	0.9539
mmlu_business_ethics	0.7900
mmlu_clinical_knowledge	0.8679
mmlu_college_biology	0.9583
mmlu_college_chemistry	0.6000
mmlu_college_computer_science	0.8400
mmlu_college_mathematics	0.7100
mmlu_college_medicine	0.8381
mmlu_college_physics	0.7255
mmlu_computer_security	0.8500
mmlu_conceptual_physics	0.9404
mmlu_econometrics	0.7632
mmlu_electrical_engineering	0.8000
mmlu_elementary_mathematics	0.8598
mmlu_formal_logic	0.7698
mmlu_global_facts	0.6000
mmlu_high_school_biology	0.9548
mmlu_high_school_chemistry	0.8227
mmlu_high_school_computer_science	0.9300
mmlu_high_school_european_history	0.8788
mmlu_high_school_geography	0.9343
mmlu_high_school_government_and_politics	0.8912
mmlu_high_school_macroeconomics	0.9385
mmlu_high_school_mathematics	0.6259
mmlu_high_school_microeconomics	0.9202
mmlu_high_school_physics	0.8278
mmlu_high_school_psychology	0.9101
mmlu_high_school_statistics	0.8657
mmlu_high_school_us_history	0.9118
mmlu_high_school_world_history	0.9494
mmlu_human_aging	0.8520
mmlu_human_sexuality	0.8779
mmlu_humanities	0.7739
mmlu_international_law	0.9091
mmlu_jurisprudence	0.8519
mmlu_logical_fallacies	0.9325
mmlu_machine_learning	0.7589
mmlu_management	0.8252
mmlu_marketing	0.8846
mmlu_medical_genetics	0.9000
mmlu_miscellaneous	0.8889
mmlu_moral_disputes	0.6936
mmlu_moral_scenarios	0.7374
mmlu_nutrition	0.9150
mmlu_other	0.8359
mmlu_philosophy	0.6817
mmlu_prehistory	0.8796
mmlu_professional_accounting	0.7837
mmlu_professional_law	0.7060
mmlu_professional_medicine	0.8125
mmlu_professional_psychology	0.8693
mmlu_public_relations	0.7909
mmlu_security_studies	0.8122
mmlu_social_sciences	0.8879
mmlu_sociology	0.9055
mmlu_stem	0.8338
mmlu_us_foreign_policy	0.9200
mmlu_virology	0.5602
mmlu_world_religions	0.9006
piqa	0.8161

How to Use

HF Usage

Step 1: Install AutoRound

pip install auto-round

Step 2: Load and run the quantized model

python

from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Qwen-Image-Bench-AutoRound-W4A16-Tuning"

# 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

bash

vllm serve Qwen-Image-Bench-AutoRound-W4A16-Tuning \
    --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.

LeaderboardModel1/Qwen-Image-Bench-AutoRound-W4A16-Tuning

Qwen-Image-Bench-AutoRound-W4A16-Tuning

Model Details