This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of apodex/Apodex-1.0-2B-SFT generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	apodex/Apodex-1.0-2B-SFT
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	2327 MB

Evaluation Results

Task	Accuracy
hellaswag	0.4535
mmlu	0.5509
mmlu_abstract_algebra	0.3300
mmlu_anatomy	0.6444
mmlu_astronomy	0.6974
mmlu_business_ethics	0.6500
mmlu_clinical_knowledge	0.5849
mmlu_college_biology	0.6250
mmlu_college_chemistry	0.4500
mmlu_college_computer_science	0.4000
mmlu_college_mathematics	0.3100
mmlu_college_medicine	0.5607
mmlu_college_physics	0.4020
mmlu_computer_security	0.7300
mmlu_conceptual_physics	0.6213
mmlu_econometrics	0.3947
mmlu_electrical_engineering	0.6000
mmlu_elementary_mathematics	0.4894
mmlu_formal_logic	0.4286
mmlu_global_facts	0.3700
mmlu_high_school_biology	0.7387
mmlu_high_school_chemistry	0.6158
mmlu_high_school_computer_science	0.6200
mmlu_high_school_european_history	0.6667
mmlu_high_school_geography	0.7172
mmlu_high_school_government_and_politics	0.6839
mmlu_high_school_macroeconomics	0.5487
mmlu_high_school_mathematics	0.3741
mmlu_high_school_microeconomics	0.6303
mmlu_high_school_physics	0.4172
mmlu_high_school_psychology	0.7431
mmlu_high_school_statistics	0.4815
mmlu_high_school_us_history	0.6373
mmlu_high_school_world_history	0.7300
mmlu_human_aging	0.5919
mmlu_human_sexuality	0.6412
mmlu_humanities	0.4803
mmlu_international_law	0.7273
mmlu_jurisprudence	0.5741
mmlu_logical_fallacies	0.6380
mmlu_machine_learning	0.4375
mmlu_management	0.6117
mmlu_marketing	0.8162
mmlu_medical_genetics	0.6500
mmlu_miscellaneous	0.6360
mmlu_moral_disputes	0.5954
mmlu_moral_scenarios	0.2469
mmlu_nutrition	0.6013
mmlu_other	0.5880
mmlu_philosophy	0.6238
mmlu_prehistory	0.5957
mmlu_professional_accounting	0.4681
mmlu_professional_law	0.3963
mmlu_professional_medicine	0.5037
mmlu_professional_psychology	0.5523
mmlu_public_relations	0.6000
mmlu_security_studies	0.6653
mmlu_social_sciences	0.6344
mmlu_sociology	0.7114
mmlu_stem	0.5382
mmlu_us_foreign_policy	0.7000
mmlu_virology	0.4277
mmlu_world_religions	0.6842
piqa	0.7171

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 = "Apodex-1.0-2B-SFT-AutoRound-W4A16-RTN"

# 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 Apodex-1.0-2B-SFT-AutoRound-W4A16-RTN \
    --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/Apodex-1.0-2B-SFT-AutoRound-W4A16-RTN

Apodex-1.0-2B-SFT-AutoRound-W4A16-RTN

Model Details