alibaba-pai / DistillQwen-ThoughtY-8B

DistillQwen-ThoughtY: Enhanced Chain-of-Thought Reasoning Models

Key Contributions

• Advanced Reasoning Models : DistillQwen-ThoughtY series (4B/8B/32B) outperform previous versions (ThoughtX) and Qwen3 in thinking mode, achieving state-of-the-art results on mathematical, scientific, and coding tasks.
• OmniThought-0528 Dataset : New 365K high-quality Chain-of-Thought (CoT) dataset distilled from DeepSeek-R1-0528 (top-tier Chinese model) with:
  • Cognitive Difficulty (CD) and Reasoning Verbosity (RV) annotations
  • Multi-teacher integration (DeepSeek-R1, DeepSeek-R1-0528, QwQ-32B)

Performance Highlights

Quick Start

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained(
    "alibaba-pai/DistillQwen-ThoughtY-4B",
    device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)

prompt = "Solve ∫x e^x dx. Show your reasoning step-by-step."
messages = [
    {"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=True # Switches between thinking and non-thinking modes. Default is True.
)
inputs = tokenizer([text], return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=32768)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Resources

• Models on Hugging Face
• OmniThought-0528 Dataset
• EasyDistill Framework

Reference

For more detailed information about the model, we encourage you to refer to our paper:

• Reasoning with OmniThought: A Large CoT Dataset with Verbosity and Cognitive Difficulty Annotations
  Wenrui Cai, Chengyu Wang, Junbing Yan, Jun Huang, Xiangzhong Fang arXiv:2505.10937

You can cite the paper using the following citation format:

@misc{cai2025reasoningomnithoughtlargecot,
      title={Reasoning with OmniThought: A Large CoT Dataset with Verbosity and Cognitive Difficulty Annotations}, 
      author={Wenrui Cai and Chengyu Wang and Junbing Yan and Jun Huang and Xiangzhong Fang},
      year={2025},
      eprint={2505.10937},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2505.10937} 
}