TinyMoE-100m-A1K-AutoRound-NVFP4-Tuning

Model Details

This model is a NVFP4 (NVIDIA FP4) quantization of FlameF0X/TinyMoE-100m-A1K generated by TUNING. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	FlameF0X/TinyMoE-100m-A1K
Quantization Tool	TUNING
Quantization Scheme	NVFP4
Quantized Size	94 MB

Evaluation Results

Task	Accuracy
hellaswag	0.2542
mmlu	0.2313
mmlu_abstract_algebra	0.2200
mmlu_anatomy	0.1926
mmlu_astronomy	0.1579
mmlu_business_ethics	0.3000
mmlu_clinical_knowledge	0.2189
mmlu_college_biology	0.2153
mmlu_college_chemistry	0.2300
mmlu_college_computer_science	0.2500
mmlu_college_mathematics	0.2000
mmlu_college_medicine	0.1850
mmlu_college_physics	0.2059
mmlu_computer_security	0.2500
mmlu_conceptual_physics	0.2553
mmlu_econometrics	0.2193
mmlu_electrical_engineering	0.2552
mmlu_elementary_mathematics	0.2090
mmlu_formal_logic	0.2381
mmlu_global_facts	0.1700
mmlu_high_school_biology	0.2097
mmlu_high_school_chemistry	0.1773
mmlu_high_school_computer_science	0.2600
mmlu_high_school_european_history	0.2000
mmlu_high_school_geography	0.1818
mmlu_high_school_government_and_politics	0.2021
mmlu_high_school_macroeconomics	0.1974
mmlu_high_school_mathematics	0.2074
mmlu_high_school_microeconomics	0.2311
mmlu_high_school_physics	0.2053
mmlu_high_school_psychology	0.2000
mmlu_high_school_statistics	0.1481
mmlu_high_school_us_history	0.2598
mmlu_high_school_world_history	0.2616
mmlu_human_aging	0.3139
mmlu_human_sexuality	0.2748
mmlu_humanities	0.2421
mmlu_international_law	0.2727
mmlu_jurisprudence	0.2593
mmlu_logical_fallacies	0.2270
mmlu_machine_learning	0.3571
mmlu_management	0.1650
mmlu_marketing	0.2949
mmlu_medical_genetics	0.2700
mmlu_miscellaneous	0.2401
mmlu_moral_disputes	0.2428
mmlu_moral_scenarios	0.2391
mmlu_nutrition	0.2157
mmlu_other	0.2417
mmlu_philosophy	0.1801
mmlu_prehistory	0.2407
mmlu_professional_accounting	0.2447
mmlu_professional_law	0.2458
mmlu_professional_medicine	0.2059
mmlu_professional_psychology	0.2582
mmlu_public_relations	0.2091
mmlu_security_studies	0.1837
mmlu_social_sciences	0.2207
mmlu_sociology	0.2438
mmlu_stem	0.2154
mmlu_us_foreign_policy	0.2700
mmlu_virology	0.3133
mmlu_world_religions	0.3158
piqa	0.5267

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 = "TinyMoE-100m-A1K-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 TinyMoE-100m-A1K-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.