Jul 2026
Post-Training Shifts Confidence: A Three-Stage Analysis of How SFT, RL, and OPD Shape CoT Calibration
A three-stage calibration framework that evaluates confidence before, during, and after chain-of-thought generation, corresponding to difficulty estimation, early termination, and answer aggregation finds that OPD provides the most useful pre-reasoning confidence, SFT gives the strongest online signal for early stoppin...
Shu-Hao Li, Guodong Du, An-Hao Zhao et al.
· arXiv.org · 1 citation