Making LLMs Say What They Think: Measuring and Improving CoT-Interpretability Alignment
arXiv:2609.38972v2 Announce Type: replace-cross Abstract: Chain-of-thought (CoT) traces often serve as a proxy for how Large Language Models (LLMs) arrive at their answers. However, growing evidence shows that models' CoT often fails to reflect their internal computations and can be changed without affecting their final answers. In this work, we measure and improve the alignment between the reasoning described in an LLM's CoT and what it computes internally. We propose CoT-Interpretability Alig…