How Chain-of-Thought Budgets Induce Overconfidence in LLMs. Investigating Calibration Drift Under Reasoning (CDUR), hypothesis lock-in mechanisms, and dynamic token budgeting via the CABStop optimal stopping algorithm.
uncertainty-estimation self-consistency ai-alignment optimal-stopping chain-of-thought autoregressive-generation inference-time-scaling expected-calibration-error verbalized-confidence compute-budget
-
Updated
Jun 16, 2026 - Python