Tess-4-27B-AutoRound-W4A16-Tuning

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of migtissera/Tess-4-27B generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	migtissera/Tess-4-27B
Quantization Tool	TUNING
Quantization Scheme	W4A16
Quantized Size	18009 MB

Evaluation Results

Task	Accuracy
hellaswag	0.2543
mmlu	0.2463
mmlu_abstract_algebra	0.3200
mmlu_anatomy	0.1852
mmlu_astronomy	0.3289
mmlu_business_ethics	0.3000
mmlu_clinical_knowledge	0.2189
mmlu_college_biology	0.2153
mmlu_college_chemistry	0.2000
mmlu_college_computer_science	0.1600
mmlu_college_mathematics	0.2900
mmlu_college_medicine	0.2081
mmlu_college_physics	0.2647
mmlu_computer_security	0.2100
mmlu_conceptual_physics	0.2596
mmlu_econometrics	0.2982
mmlu_electrical_engineering	0.2759
mmlu_elementary_mathematics	0.2116
mmlu_formal_logic	0.2302
mmlu_global_facts	0.3100
mmlu_high_school_biology	0.2290
mmlu_high_school_chemistry	0.2611
mmlu_high_school_computer_science	0.2500
mmlu_high_school_european_history	0.2364
mmlu_high_school_geography	0.2424
mmlu_high_school_government_and_politics	0.2124
mmlu_high_school_macroeconomics	0.2154
mmlu_high_school_mathematics	0.2444
mmlu_high_school_microeconomics	0.2185
mmlu_high_school_physics	0.2583
mmlu_high_school_psychology	0.2165
mmlu_high_school_statistics	0.1759
mmlu_high_school_us_history	0.2304
mmlu_high_school_world_history	0.2447
mmlu_human_aging	0.2780
mmlu_human_sexuality	0.2137
mmlu_humanities	0.2495
mmlu_international_law	0.2727
mmlu_jurisprudence	0.2685
mmlu_logical_fallacies	0.2086
mmlu_machine_learning	0.2500
mmlu_management	0.1553
mmlu_marketing	0.2778
mmlu_medical_genetics	0.2300
mmlu_miscellaneous	0.2682
mmlu_moral_disputes	0.2370
mmlu_moral_scenarios	0.2525
mmlu_nutrition	0.2516
mmlu_other	0.2572
mmlu_philosophy	0.2379
mmlu_prehistory	0.2778
mmlu_professional_accounting	0.2447
mmlu_professional_law	0.2549
mmlu_professional_medicine	0.2463
mmlu_professional_psychology	0.2614
mmlu_public_relations	0.2818
mmlu_security_studies	0.2367
mmlu_social_sciences	0.2385
mmlu_sociology	0.2637
mmlu_stem	0.2385
mmlu_us_foreign_policy	0.2700
mmlu_virology	0.3313
mmlu_world_religions	0.2456
piqa	0.5299

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 = "Tess-4-27B-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 Tess-4-27B-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.