DeepSeek-R1-Distill-Qwen-1.5B-AutoRound-NVFP4-Tuning

Model Details

This model is a NVFP4 (NVIDIA FP4) quantization of deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
Quantization Tool	TUNING
Quantization Scheme	NVFP4
Quantized Size	1805 MB

Evaluation Results

Task	Accuracy
hellaswag	0.3582
mmlu	0.3565
mmlu_abstract_algebra	0.3100
mmlu_anatomy	0.3481
mmlu_astronomy	0.3421
mmlu_business_ethics	0.4000
mmlu_clinical_knowledge	0.3358
mmlu_college_biology	0.3264
mmlu_college_chemistry	0.3500
mmlu_college_computer_science	0.3900
mmlu_college_mathematics	0.3800
mmlu_college_medicine	0.3526
mmlu_college_physics	0.2745
mmlu_computer_security	0.3400
mmlu_conceptual_physics	0.4298
mmlu_econometrics	0.2719
mmlu_electrical_engineering	0.4069
mmlu_elementary_mathematics	0.4392
mmlu_formal_logic	0.3968
mmlu_global_facts	0.3400
mmlu_high_school_biology	0.4032
mmlu_high_school_chemistry	0.3744
mmlu_high_school_computer_science	0.4400
mmlu_high_school_european_history	0.3576
mmlu_high_school_geography	0.3586
mmlu_high_school_government_and_politics	0.3316
mmlu_high_school_macroeconomics	0.3590
mmlu_high_school_mathematics	0.3000
mmlu_high_school_microeconomics	0.4412
mmlu_high_school_physics	0.2649
mmlu_high_school_psychology	0.4330
mmlu_high_school_statistics	0.3657
mmlu_high_school_us_history	0.3039
mmlu_high_school_world_history	0.3713
mmlu_human_aging	0.4126
mmlu_human_sexuality	0.4198
mmlu_humanities	0.3105
mmlu_international_law	0.4298
mmlu_jurisprudence	0.4630
mmlu_logical_fallacies	0.3926
mmlu_machine_learning	0.2143
mmlu_management	0.4563
mmlu_marketing	0.5897
mmlu_medical_genetics	0.4000
mmlu_miscellaneous	0.4125
mmlu_moral_disputes	0.3671
mmlu_moral_scenarios	0.2324
mmlu_nutrition	0.4150
mmlu_other	0.3911
mmlu_philosophy	0.3923
mmlu_prehistory	0.3488
mmlu_professional_accounting	0.2589
mmlu_professional_law	0.2725
mmlu_professional_medicine	0.2978
mmlu_professional_psychology	0.3203
mmlu_public_relations	0.4455
mmlu_security_studies	0.3878
mmlu_social_sciences	0.3848
mmlu_sociology	0.4677
mmlu_stem	0.3635
mmlu_us_foreign_policy	0.4800
mmlu_virology	0.4217
mmlu_world_religions	0.2807
piqa	0.6485

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 = "DeepSeek-R1-Distill-Qwen-1.5B-AutoRound-NVFP4-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 DeepSeek-R1-Distill-Qwen-1.5B-AutoRound-NVFP4-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.