G9v3-39A5B-AutoRound-W4A16-Tuning

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of ai9stars/G9v3-39A5B generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	ai9stars/G9v3-39A5B
Quantization Tool	TUNING
Quantization Scheme	W4A16
Quantized Size	20109 MB

Evaluation Results

Task	Accuracy
hellaswag	0.5974
mmlu	0.7574
mmlu_abstract_algebra	0.5300
mmlu_anatomy	0.7704
mmlu_astronomy	0.9013
mmlu_business_ethics	0.7800
mmlu_clinical_knowledge	0.8453
mmlu_college_biology	0.9028
mmlu_college_chemistry	0.5400
mmlu_college_computer_science	0.7200
mmlu_college_mathematics	0.5800
mmlu_college_medicine	0.7803
mmlu_college_physics	0.5196
mmlu_computer_security	0.8100
mmlu_conceptual_physics	0.8468
mmlu_econometrics	0.6491
mmlu_electrical_engineering	0.8069
mmlu_elementary_mathematics	0.6640
mmlu_formal_logic	0.5873
mmlu_global_facts	0.4900
mmlu_high_school_biology	0.9129
mmlu_high_school_chemistry	0.7044
mmlu_high_school_computer_science	0.8600
mmlu_high_school_european_history	0.8485
mmlu_high_school_geography	0.9091
mmlu_high_school_government_and_politics	0.9534
mmlu_high_school_macroeconomics	0.8179
mmlu_high_school_mathematics	0.4593
mmlu_high_school_microeconomics	0.9118
mmlu_high_school_physics	0.6225
mmlu_high_school_psychology	0.9138
mmlu_high_school_statistics	0.7315
mmlu_high_school_us_history	0.9265
mmlu_high_school_world_history	0.8819
mmlu_human_aging	0.7848
mmlu_human_sexuality	0.8779
mmlu_humanities	0.6812
mmlu_international_law	0.9091
mmlu_jurisprudence	0.8796
mmlu_logical_fallacies	0.8650
mmlu_machine_learning	0.6250
mmlu_management	0.8932
mmlu_marketing	0.9231
mmlu_medical_genetics	0.8900
mmlu_miscellaneous	0.8991
mmlu_moral_disputes	0.8092
mmlu_moral_scenarios	0.4257
mmlu_nutrition	0.8693
mmlu_other	0.8120
mmlu_philosophy	0.8167
mmlu_prehistory	0.8704
mmlu_professional_accounting	0.5603
mmlu_professional_law	0.5874
mmlu_professional_medicine	0.8971
mmlu_professional_psychology	0.8284
mmlu_public_relations	0.7727
mmlu_security_studies	0.7755
mmlu_social_sciences	0.8580
mmlu_sociology	0.8955
mmlu_stem	0.7190
mmlu_us_foreign_policy	0.9100
mmlu_virology	0.5602
mmlu_world_religions	0.8713
piqa	0.8009

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 = "G9v3-39A5B-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

vllm serve G9v3-39A5B-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:

• 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.