Qwen3 Technical Report

Team, 2025, Qwen3 Technical Report (arXiv preprint arXiv:2505.09388)

Subscriber deck — free during early access

1 cards in “4.1 Long-CoT Cold Start” · whole deck

4.1 Long-CoT Cold Start

  1. How was Qwen3’s cold-start reasoning data filtered, and why keep that stage small?
    Queries had to be verifiable and need reasoning (dropped if Qwen2.5-72B-Instruct answered them without chain of thought); QwQ-32B responses were removed if wrong, repetitive, guessed, inconsistent with their summary or language-mixed. The stage uses few samples and steps to instill reasoning patterns without limiting what RL can later achieve.

Source