Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-AutoRound-W4A16-Tuning

Model Details

This model is a int4 weight-only quantization with group_size 128 and symmetric quantization of nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451 generated by AutoRound. Please follow the license of the original model.

Quantization Details

Attribute	Value
Base Model	nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451
Quantization Tool	AutoRound
Quantization Scheme	W4A16
Quantized Size	18009 MB

Evaluation Results

Task	Accuracy
hellaswag	0.6569
mmlu	0.8512
mmlu_abstract_algebra	0.7700
mmlu_anatomy	0.8444
mmlu_astronomy	0.9342
mmlu_business_ethics	0.8200
mmlu_clinical_knowledge	0.9057
mmlu_college_biology	0.9653
mmlu_college_chemistry	0.6800
mmlu_college_computer_science	0.8400
mmlu_college_mathematics	0.7100
mmlu_college_medicine	0.8844
mmlu_college_physics	0.7059
mmlu_computer_security	0.8700
mmlu_conceptual_physics	0.9447
mmlu_econometrics	0.8158
mmlu_electrical_engineering	0.8345
mmlu_elementary_mathematics	0.8598
mmlu_formal_logic	0.7460
mmlu_global_facts	0.5700
mmlu_high_school_biology	0.9581
mmlu_high_school_chemistry	0.8325
mmlu_high_school_computer_science	0.9200
mmlu_high_school_european_history	0.9030
mmlu_high_school_geography	0.9444
mmlu_high_school_government_and_politics	0.9896
mmlu_high_school_macroeconomics	0.9359
mmlu_high_school_mathematics	0.6407
mmlu_high_school_microeconomics	0.9538
mmlu_high_school_physics	0.8477
mmlu_high_school_psychology	0.9560
mmlu_high_school_statistics	0.9028
mmlu_high_school_us_history	0.9265
mmlu_high_school_world_history	0.9494
mmlu_human_aging	0.8475
mmlu_human_sexuality	0.9160
mmlu_humanities	0.7968
mmlu_international_law	0.9256
mmlu_jurisprudence	0.8889
mmlu_logical_fallacies	0.9202
mmlu_machine_learning	0.7321
mmlu_management	0.8835
mmlu_marketing	0.9573
mmlu_medical_genetics	0.9700
mmlu_miscellaneous	0.9438
mmlu_moral_disputes	0.8439
mmlu_moral_scenarios	0.7140
mmlu_nutrition	0.9118
mmlu_other	0.8806
mmlu_philosophy	0.8650
mmlu_prehistory	0.9105
mmlu_professional_accounting	0.8156
mmlu_professional_law	0.7073
mmlu_professional_medicine	0.9596
mmlu_professional_psychology	0.8791
mmlu_public_relations	0.7636
mmlu_security_studies	0.8204
mmlu_social_sciences	0.9129
mmlu_sociology	0.9403
mmlu_stem	0.8430
mmlu_us_foreign_policy	0.9300
mmlu_virology	0.5663
mmlu_world_religions	0.9006
piqa	0.8172

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 = "Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-AutoRound-W4A16-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 Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-AutoRound-W4A16-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:

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.