Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought 事件

Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought arXiv:2603.05488v4 Announce Type: replace Abstract: We provide evidence of performative chain-of-thought (CoT) in reasoning models, where a model becomes strongly confident in its final answer, but continues generating tokens without revealing its internal belief. Our analysis compares activation probing, early forced answering, and a CoT monitor across two large models (DeepSeek-R1 671B & GPT-OSS 120B) and find task difficult