Qwen3 Technical Report
Team, 2025, Qwen3 Technical Report (arXiv preprint arXiv:2505.09388)
Subscriber deck — free during early access
1 cards in “4.4 General RL” · whole deck
4.4 General RL
- What rewards did Qwen3’s general RL stage use?Over 20 tasks (instruction following, format and mode-flag following, preference alignment, tool use in real multi-turn environments, RAG and more), scored by rule-based rewards, by Qwen2.5-72B-Instruct judging against a reference answer, or by a reward model trained on human preferences when no reference exists.
- Team, 2025, Qwen3 Technical Report (arXiv preprint arXiv:2505.09388), §4.4
Source
- Qwen Team. 2025. Qwen3 Technical Report. In arXiv preprint arXiv:2505.09388. https://arxiv.org/abs/2505.09388