Analyzing and Improving Chain-of-Thought Monitorability Through Information Theory
arXiv:2602.18297v1 Announce Type: cross Abstract: Chain-of-thought (CoT) monitors are LLM-based systems that analyze reasoning traces to detect when outputs may exhibit attributes of interest, such …
Usman Anwar, Tim Bakker, Dana Kianfar, Cristina Pinneri, Christos Louizos
7 views