arXiv cs.CL
· Papers
Measuring and Mitigating Post-hoc Rationalization in Reverse Chain-of-Thought Generation
arXiv:2602.14469v4 Announce Type: replace Abstract: Reverse Chain-of-Thought Generation (RCG) synthesizes reasoning traces from query-answer pairs, but answer-visible generation can justify a pre-committed answer rather than derive it. This post-hoc rationalization creates a train-inference mismatch because student mod