This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of kai-os/Grug-35B-A3B generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	kai-os/Grug-35B-A3B
Quantization Tool	TUNING
Quantization Scheme	W4A16
Quantized Size	18653 MB

Evaluation Results

Task	Accuracy
hellaswag	0.6252
mmlu	0.8287
mmlu_abstract_algebra	0.6900
mmlu_anatomy	0.8593
mmlu_astronomy	0.9276
mmlu_business_ethics	0.8500
mmlu_clinical_knowledge	0.8868
mmlu_college_biology	0.9375
mmlu_college_chemistry	0.6400
mmlu_college_computer_science	0.7400
mmlu_college_mathematics	0.6900
mmlu_college_medicine	0.8439
mmlu_college_physics	0.6667
mmlu_computer_security	0.8700
mmlu_conceptual_physics	0.9319
mmlu_econometrics	0.7895
mmlu_electrical_engineering	0.8276
mmlu_elementary_mathematics	0.8201
mmlu_formal_logic	0.6905
mmlu_global_facts	0.5800
mmlu_high_school_biology	0.9613
mmlu_high_school_chemistry	0.8079
mmlu_high_school_computer_science	0.9100
mmlu_high_school_european_history	0.8727
mmlu_high_school_geography	0.9343
mmlu_high_school_government_and_politics	0.9845
mmlu_high_school_macroeconomics	0.8923
mmlu_high_school_mathematics	0.6185
mmlu_high_school_microeconomics	0.9664
mmlu_high_school_physics	0.8278
mmlu_high_school_psychology	0.9523
mmlu_high_school_statistics	0.8056
mmlu_high_school_us_history	0.9118
mmlu_high_school_world_history	0.9030
mmlu_human_aging	0.8386
mmlu_human_sexuality	0.8931
mmlu_humanities	0.7624
mmlu_international_law	0.9339
mmlu_jurisprudence	0.8981
mmlu_logical_fallacies	0.9264
mmlu_machine_learning	0.7679
mmlu_management	0.8932
mmlu_marketing	0.9444
mmlu_medical_genetics	0.9600
mmlu_miscellaneous	0.9374
mmlu_moral_disputes	0.8555
mmlu_moral_scenarios	0.6000
mmlu_nutrition	0.9020
mmlu_other	0.8664
mmlu_philosophy	0.8617
mmlu_prehistory	0.8981
mmlu_professional_accounting	0.7376
mmlu_professional_law	0.6838
mmlu_professional_medicine	0.9375
mmlu_professional_psychology	0.8725
mmlu_public_relations	0.7455
mmlu_security_studies	0.8367
mmlu_social_sciences	0.9035
mmlu_sociology	0.9303
mmlu_stem	0.8173
mmlu_us_foreign_policy	0.9300
mmlu_virology	0.5964
mmlu_world_religions	0.9006
piqa	0.8183

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 = "Grug-35B-A3B-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 Grug-35B-A3B-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/Grug-35B-A3B-AutoRound-W4A16-Tuning

Grug-35B-A3B-AutoRound-W4A16-Tuning

Model Details