用特定标记符实现无需训练的推理预算控制，提升模型表现
Mid-Think: Training-Free Intermediate-Budget Reasoning via Token-Level Triggers
- 通过分析触发词发现'Okay'和'</think>换行'是推理开关的关键
- 提出Mid-Think格式，在准确率与长度间取得更好平衡
- 适用于推理阶段调控和强化学习训练，显著提效增质
混合推理语言模型通常通过高阶的Think/No-think指令控制推理行为，但我们发现这种模式切换主要由少数触发词驱动，而非指令本身。通过注意力分析和受控提示实验，我们发现'Okay'这一词元会诱导推理行为，而'</think>'后的换行符则抑制该行为。基于此，我们提出Mid-Think——一种无需训练的提示格式，结合这些触发符实现中等预算推理，在准确率-长度权衡上持续优于固定词元和提示基线。此外，将Mid-Think应用于SFT后的强化学习训练，使Qwen3-8B在AIME上的性能从69.8%提升至72.4%，在GPQA上从58.5%提升至61.1%，训练时间减少约15%，验证了其在推理控制和基于强化学习的推理训练中的有效性。
原文摘要 · Abstract (English)
Hybrid reasoning language models are commonly controlled through high-level Think/No-think instructions to regulate reasoning behavior, yet we found that such mode switching is largely driven by a small set of trigger tokens rather than the instructions themselves. Through attention analysis and controlled prompting experiments, we show that a leading ``Okay'' token induces reasoning behavior, while the newline pattern following ``</think>'' suppresses it. Based on this observation, we propose Mid-Think, a simple training-free prompting format that combines these triggers to achieve intermediate-budget reasoning, consistently outperforming fixed-token and prompt-based baselines in terms of the accuracy-length trade-off. Furthermore, applying Mid-Think to RL training after SFT reduces training time by approximately 15% while improving final performance of Qwen3-8B on AIME from 69.8% to 72.4% and on GPQA from 58.5% to 61.1%, demonstrating its effectiveness for both inference-time control and RL-based reasoning training.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。