North-Mini-Code-1.0-AutoRound-W4A16-Tuning

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of CohereLabs/North-Mini-Code-1.0 generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	CohereLabs/North-Mini-Code-1.0
Quantization Tool	TUNING
Quantization Scheme	W4A16
Quantized Size	15886 MB

Evaluation Results

Task	Accuracy
hellaswag	0.5638
mmlu	0.7356
mmlu_abstract_algebra	0.6500
mmlu_anatomy	0.7778
mmlu_astronomy	0.8882
mmlu_business_ethics	0.7800
mmlu_clinical_knowledge	0.8226
mmlu_college_biology	0.8819
mmlu_college_chemistry	0.5500
mmlu_college_computer_science	0.6600
mmlu_college_mathematics	0.5300
mmlu_college_medicine	0.7225
mmlu_college_physics	0.5294
mmlu_computer_security	0.8300
mmlu_conceptual_physics	0.8340
mmlu_econometrics	0.6842
mmlu_electrical_engineering	0.8276
mmlu_elementary_mathematics	0.6905
mmlu_formal_logic	0.6190
mmlu_global_facts	0.4900
mmlu_high_school_biology	0.8903
mmlu_high_school_chemistry	0.6897
mmlu_high_school_computer_science	0.8400
mmlu_high_school_european_history	0.7576
mmlu_high_school_geography	0.8788
mmlu_high_school_government_and_politics	0.9430
mmlu_high_school_macroeconomics	0.8308
mmlu_high_school_mathematics	0.4148
mmlu_high_school_microeconomics	0.9328
mmlu_high_school_physics	0.6291
mmlu_high_school_psychology	0.9193
mmlu_high_school_statistics	0.6944
mmlu_high_school_us_history	0.8725
mmlu_high_school_world_history	0.8565
mmlu_human_aging	0.7578
mmlu_human_sexuality	0.8321
mmlu_humanities	0.6444
mmlu_international_law	0.8595
mmlu_jurisprudence	0.8611
mmlu_logical_fallacies	0.8160
mmlu_machine_learning	0.6071
mmlu_management	0.8447
mmlu_marketing	0.9060
mmlu_medical_genetics	0.8700
mmlu_miscellaneous	0.8582
mmlu_moral_disputes	0.7977
mmlu_moral_scenarios	0.3553
mmlu_nutrition	0.8333
mmlu_other	0.7799
mmlu_philosophy	0.8103
mmlu_prehistory	0.8951
mmlu_professional_accounting	0.5780
mmlu_professional_law	0.5417
mmlu_professional_medicine	0.8088
mmlu_professional_psychology	0.8088
mmlu_public_relations	0.7273
mmlu_security_studies	0.8000
mmlu_social_sciences	0.8544
mmlu_sociology	0.8806
mmlu_stem	0.7120
mmlu_us_foreign_policy	0.9100
mmlu_virology	0.5301
mmlu_world_religions	0.8830
piqa	0.7764

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 = "North-Mini-Code-1.0-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 North-Mini-Code-1.0-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.