Qwen3.5-35B-A3B-heretic-v2-AutoRound-W4A16-RTN

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of llmfan46/Qwen3.5-35B-A3B-heretic-v2 generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	llmfan46/Qwen3.5-35B-A3B-heretic-v2
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	19504 MB

Evaluation Results

Task	Accuracy
hellaswag	0.6198
mmlu	0.8064
mmlu_abstract_algebra	0.6500
mmlu_anatomy	0.8370
mmlu_astronomy	0.9211
mmlu_business_ethics	0.8500
mmlu_clinical_knowledge	0.9019
mmlu_college_biology	0.9514
mmlu_college_chemistry	0.6400
mmlu_college_computer_science	0.7500
mmlu_college_mathematics	0.6800
mmlu_college_medicine	0.8555
mmlu_college_physics	0.7059
mmlu_computer_security	0.8300
mmlu_conceptual_physics	0.9404
mmlu_econometrics	0.7368
mmlu_electrical_engineering	0.8345
mmlu_elementary_mathematics	0.7937
mmlu_formal_logic	0.7063
mmlu_global_facts	0.5100
mmlu_high_school_biology	0.9516
mmlu_high_school_chemistry	0.7980
mmlu_high_school_computer_science	0.8900
mmlu_high_school_european_history	0.8242
mmlu_high_school_geography	0.9495
mmlu_high_school_government_and_politics	0.9793
mmlu_high_school_macroeconomics	0.8949
mmlu_high_school_mathematics	0.5519
mmlu_high_school_microeconomics	0.9664
mmlu_high_school_physics	0.7748
mmlu_high_school_psychology	0.9486
mmlu_high_school_statistics	0.8009
mmlu_high_school_us_history	0.9118
mmlu_high_school_world_history	0.9198
mmlu_human_aging	0.7937
mmlu_human_sexuality	0.8855
mmlu_humanities	0.7133
mmlu_international_law	0.9091
mmlu_jurisprudence	0.8889
mmlu_logical_fallacies	0.8834
mmlu_machine_learning	0.7679
mmlu_management	0.9223
mmlu_marketing	0.9402
mmlu_medical_genetics	0.9500
mmlu_miscellaneous	0.9374
mmlu_moral_disputes	0.8468
mmlu_moral_scenarios	0.4156
mmlu_nutrition	0.8889
mmlu_other	0.8590
mmlu_philosophy	0.8521
mmlu_prehistory	0.8920
mmlu_professional_accounting	0.7305
mmlu_professional_law	0.6552
mmlu_professional_medicine	0.9375
mmlu_professional_psychology	0.8791
mmlu_public_relations	0.7455
mmlu_security_studies	0.8122
mmlu_social_sciences	0.8999
mmlu_sociology	0.9204
mmlu_stem	0.8024
mmlu_us_foreign_policy	0.9200
mmlu_virology	0.5542
mmlu_world_religions	0.8947
piqa	0.8183

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 = "Qwen3.5-35B-A3B-heretic-v2-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

vllm serve Qwen3.5-35B-A3B-heretic-v2-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.