通过隐状态突变检测，实时纠正大模型推理中的思维僵化问题。
Mitigating Cognitive Inertia in Large Reasoning Models via Latent Spike Steering
- 检测隐藏层中突增的L2距离，定位推理转折点。
- 在多个基准上减少冗余循环，准确率提升显著。
- 无需微调，适合部署于现有推理系统中。
尽管大型推理模型（LRMs）通过扩大推理时计算量取得了显著性能，但常出现认知惯性问题，表现为过度思考（运动惯性）或思维僵化（方向惯性）。现有检测方法多依赖文本层面的启发式规则（如自我修正标记），难以捕捉模型内部未表达的冲突。为此，我们提出STARS（Spike-Triggered Adaptive Reasoning Steering）——一种无需训练的框架，通过监控隐状态动态来纠正认知惯性。STARS通过识别隐藏状态中显著的L2距离突变，定位关键推理转折点（认知转折点），并利用几何轨迹分析判断转变结构，实时注入状态感知的语言提示以引导模型。实验结果表明，STARS能有效抑制冗余循环，并通过自适应修正错误推理路径提升准确性。该方法为不需额外微调的大型推理模型提供了一种鲁棒、无监督的推理优化机制。
原文摘要 · Abstract (English)
While Large Reasoning Models (LRMs) have achieved remarkable performance by scaling test-time compute, they frequently suffer from Cognitive Inertia, a failure pattern manifesting as either overthinking (inertia of motion) or reasoning rigidity (inertia of direction). Existing detection methods, typically relying on superficial textual heuristics like self-correction tokens, often fail to capture the model's unvoiced internal conflicts. To address this, we propose STARS (Spike-Triggered Adaptive Reasoning Steering), a training-free framework designed to rectify cognitive inertia by monitoring latent dynamics. STARS identifies Cognitive Pivots-critical moments of reasoning transition-by detecting distinct L2 distance spikes in the hidden states. Upon detection, the framework employs geometric trajectory analysis to diagnose the structural nature of the transition and injects state-aware language cues to steer the model in real-time. Our experiments across diverse benchmarks confirm that STARS efficiently curtails redundant loops while improving accuracy through the adaptive correction of erroneous trajectories. STARS offers a robust, unsupervised mechanism to optimize the reasoning process of LRMs without requiring additional fine-tuning.
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。