思维泄漏：对混合推理模型中NoThink后训练的因果审计

中文摘要 AI 辅助

在NoThink模式下对混合推理模型进行后训练，作为一种在保持推理速度的同时提升性能的方法，已引起越来越多的关注。然而，这些收益可能依赖于通过基础模型的Think模式已可获得的思维行为。我们在因果中介框架中阐述了这种思维泄漏，并沿一条简单的基于基础模型的激活方向，通过双向干预来审计其贡献。在竞争性数学基准上，针对三个模型和三种后训练方法，我们发现泄漏是真实的、因果性的且显著的：行为和表征分析揭示了向Think模式的转变，沿该方向引导基础模型可复现后训练的大部分准确率提升，而反向引导一个检查点则会消除其所得收益的相当大一部分。在九个具有正向NoThink收益的对齐检查点中，由此产生的泄漏比率介于42%至79%之间。这些干预支持了思维泄漏具有实质性因果贡献的观点。我们的研究结果表明，一种后训练方法的表面优势可能因此反映了向Think模式的更大漂移，从而掩盖了它是在NoThink模式内提升了能力，还是更有效地重新调用了已有的Think行为。

英文摘要

Post-training hybrid reasoning models in NoThink mode has attracted growing interest as a way to improve performance while keeping inference fast. However, these gains may draw on thinking behavior already accessible through the base model's Think mode. We formulate this thinking leakage in a causal mediation framework and audit its contribution using bidirectional interventions along a simple base-derived activation direction. Across three models and three post-training methods on competition math benchmarks, we find that leakage is real, causal, and substantial: behavioral and representational analyses reveal shifts toward Think, steering the base model along this direction reproduces most of the post-training accuracy gain, and counter-steering a checkpoint removes a substantial share of what it gains. Across nine aligned checkpoints with positive NoThink gains, the resulting leakage ratio ranges from 42% to 79%. These interventions support a substantial causal contribution of thinking leakage. Our findings show that a post-training method's apparent advantage can therefore reflect greater drift toward Think, obscuring whether it improves capability within NoThink or more effectively re-invokes existing Think behavior.

发表机构

• College of Information Sciences and Technology（信息科学与技术学院）
• Pennsylvania State University（宾夕法尼亚州立大学）

机构由 AI 辅助整理，请以论文原文为准。