LFM2-Research-AutoRound-W4A16-Tuning

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of FlameF0X/LFM2-Research generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	FlameF0X/LFM2-Research
Quantization Tool	TUNING
Quantization Scheme	W4A16
Quantized Size	62 MB

Evaluation Results

Task	Accuracy
hellaswag	0.2633
mmlu	0.2418
mmlu_abstract_algebra	0.2100
mmlu_anatomy	0.2000
mmlu_astronomy	0.2434
mmlu_business_ethics	0.2500
mmlu_clinical_knowledge	0.2528
mmlu_college_biology	0.2569
mmlu_college_chemistry	0.2400
mmlu_college_computer_science	0.2400
mmlu_college_mathematics	0.3000
mmlu_college_medicine	0.2486
mmlu_college_physics	0.1275
mmlu_computer_security	0.2700
mmlu_conceptual_physics	0.3277
mmlu_econometrics	0.2544
mmlu_electrical_engineering	0.1793
mmlu_elementary_mathematics	0.2275
mmlu_formal_logic	0.2143
mmlu_global_facts	0.3500
mmlu_high_school_biology	0.2194
mmlu_high_school_chemistry	0.2414
mmlu_high_school_computer_science	0.3100
mmlu_high_school_european_history	0.2303
mmlu_high_school_geography	0.2222
mmlu_high_school_government_and_politics	0.2073
mmlu_high_school_macroeconomics	0.2103
mmlu_high_school_mathematics	0.2407
mmlu_high_school_microeconomics	0.2437
mmlu_high_school_physics	0.2715
mmlu_high_school_psychology	0.1963
mmlu_high_school_statistics	0.2361
mmlu_high_school_us_history	0.2598
mmlu_high_school_world_history	0.2658
mmlu_human_aging	0.3094
mmlu_human_sexuality	0.2214
mmlu_humanities	0.2453
mmlu_international_law	0.2893
mmlu_jurisprudence	0.2778
mmlu_logical_fallacies	0.2086
mmlu_machine_learning	0.2589
mmlu_management	0.2233
mmlu_marketing	0.2265
mmlu_medical_genetics	0.2900
mmlu_miscellaneous	0.2580
mmlu_moral_disputes	0.2514
mmlu_moral_scenarios	0.2246
mmlu_nutrition	0.2059
mmlu_other	0.2459
mmlu_philosophy	0.2605
mmlu_prehistory	0.2438
mmlu_professional_accounting	0.2411
mmlu_professional_law	0.2432
mmlu_professional_medicine	0.1618
mmlu_professional_psychology	0.2859
mmlu_public_relations	0.2000
mmlu_security_studies	0.2245
mmlu_social_sciences	0.2320
mmlu_sociology	0.2289
mmlu_stem	0.2420
mmlu_us_foreign_policy	0.2700
mmlu_virology	0.2590
mmlu_world_religions	0.3099
piqa	0.5158

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 = "LFM2-Research-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 LFM2-Research-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.