[2509.23392] Your Models Have Thought Enough: Training Large Reasoning Models to Stop Overthinking

Computer Science > Artificial Intelligence arXiv:2509.23392 (cs) [Submitted on 27 Sep 2025 (v1), last revised 30 Mar 2026 (this version, v3)] Title:Your Models Have Thought Enough: Training Large Reasoning Models to Stop Overthinking Authors:Jinyi Han, Ying Huang, Ying Liao, Zishang Jiang, Xikun Lu, Haiquan Zhao, Xinyi Wang, Guanghao Zhou, Sihang Jiang, Jiaqing Liang, Weikang Zhou, Zeye Sun, Fei Yu, Yanghua Xiao View a PDF of the paper titled Your Models Have Thought Enough: Training Large Reasoning Models to Stop Overthinking, by Jinyi Han and 13 other authors View PDF HTML (experimental) Abstract:Large Reasoning Models (LRMs) have achieved impressive performance on challenging tasks, yet their deep reasoning often incurs substantial computational costs. To achieve efficient reasoning, existing reinforcement learning methods still struggle to construct short reasoning path during the rollout stage, limiting effective learning. Inspired by Evidence Accumulation Models, we find that LRMs have accumulated sufficient information early in reasoning, making further reasoning steps redundant. Based on this insight, we propose Just-Enough Thinking (JET), which trains models to proactively terminate unnecessary reasoning. JET performs trajectory truncation during rollout to expose the model to short, distributionally consistent reasoning paths. Besides, it uses a quality-controlled length reward to better encourage concise reasoning while maintaining correctness. Extensive experim...