This model is a NVFP4 (NVIDIA FP4) quantization of sasa2000/Qwen3-Swallow-30B-A3B-RL-v0.2-heretic generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	sasa2000/Qwen3-Swallow-30B-A3B-RL-v0.2-heretic
Quantization Tool	AutoRound
Quantization Scheme	NVFP4
Quantized Size	18500 MB

Evaluation Results

Task	Accuracy
hellaswag	0.5852
mmlu	0.7504
mmlu_abstract_algebra	0.6400
mmlu_anatomy	0.7926
mmlu_astronomy	0.8947
mmlu_business_ethics	0.8000
mmlu_clinical_knowledge	0.8264
mmlu_college_biology	0.8958
mmlu_college_chemistry	0.5600
mmlu_college_computer_science	0.7700
mmlu_college_mathematics	0.6100
mmlu_college_medicine	0.7514
mmlu_college_physics	0.6569
mmlu_computer_security	0.8600
mmlu_conceptual_physics	0.8936
mmlu_econometrics	0.6667
mmlu_electrical_engineering	0.8000
mmlu_elementary_mathematics	0.8280
mmlu_formal_logic	0.6032
mmlu_global_facts	0.4400
mmlu_high_school_biology	0.9323
mmlu_high_school_chemistry	0.7291
mmlu_high_school_computer_science	0.8500
mmlu_high_school_european_history	0.8061
mmlu_high_school_geography	0.8889
mmlu_high_school_government_and_politics	0.9482
mmlu_high_school_macroeconomics	0.8667
mmlu_high_school_mathematics	0.5667
mmlu_high_school_microeconomics	0.9496
mmlu_high_school_physics	0.7219
mmlu_high_school_psychology	0.9266
mmlu_high_school_statistics	0.7454
mmlu_high_school_us_history	0.8529
mmlu_high_school_world_history	0.8692
mmlu_human_aging	0.7668
mmlu_human_sexuality	0.8397
mmlu_humanities	0.6295
mmlu_international_law	0.8430
mmlu_jurisprudence	0.8333
mmlu_logical_fallacies	0.8344
mmlu_machine_learning	0.6518
mmlu_management	0.8252
mmlu_marketing	0.9274
mmlu_medical_genetics	0.8800
mmlu_miscellaneous	0.8889
mmlu_moral_disputes	0.7543
mmlu_moral_scenarios	0.3341
mmlu_nutrition	0.8562
mmlu_other	0.8011
mmlu_philosophy	0.7846
mmlu_prehistory	0.8580
mmlu_professional_accounting	0.6064
mmlu_professional_law	0.5287
mmlu_professional_medicine	0.8566
mmlu_professional_psychology	0.8007
mmlu_public_relations	0.7545
mmlu_security_studies	0.7714
mmlu_social_sciences	0.8599
mmlu_sociology	0.8905
mmlu_stem	0.7739
mmlu_us_foreign_policy	0.9100
mmlu_virology	0.5602
mmlu_world_religions	0.8889
piqa	0.7911

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 = "Qwen3-Swallow-30B-A3B-RL-v0.2-heretic-AutoRound-NVFP4-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 Qwen3-Swallow-30B-A3B-RL-v0.2-heretic-AutoRound-NVFP4-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/Qwen3-Swallow-30B-A3B-RL-v0.2-heretic-AutoRound-NVFP4-RTN

Qwen3-Swallow-30B-A3B-RL-v0.2-heretic-AutoRound-NVFP4-RTN

Model Details