监测内部独白：探针轨迹揭示大型推理模型的推理动态

View PDF HTML (experimental)

Abstract:Large Reasoning Models (LRMs) introduce new opportunities for safety monitoring through their Chain-of-Thought (CoT) reasoning. However, CoT is not always faithful to the model's final output, undermining its reliability as a monitoring tool. To address this, we investigate the hidden representations of LRMs to determine whether future behavior can be predicted from prompt and CoT representations. By evaluating a probe at each generated token, we construct a probe trajectory, the continuous evolution of a concept's probability across the reasoning process. To characterize these temporal dynamics, we extract signal-processing features that capture volatility, trend, and steady-state behavior. We find that trajectory features help precisely where the static probe is weakest, adding 21.7 AUROC points on average, and improving in every such case where that probe falls below 70. We also present two methodological insights. First, template-based training data achieves near-parity with dynamically generated model responses, eliminating the need for a costly initial inference and labeling. Second, the use of a pooling operation is necessary: last-token methods achieve poor performance, whereas average and max pooling achieve 78.5% to 97.4% AUROC and yield stable probe trajectories. Across seven reasoning models from three families and six evaluation datasets, we find probe trajectories to be a promising signal for monitoring model behavior.
Warning: This article contains potentially harmful content.

Submission history

From: Maciej Chrabaszcz [view email]
[v1] Mon, 18 May 2026 15:29:04 UTC (4,382 KB)
[v2] Mon, 2 Nov 2026 08:55:39 UTC (5,809 KB)