Does Your Reasoning Model Implicitly Know When to Stop Thinking?

In three linesLong reasoning models (LRMs) generate redundant chains of thought uncorrelated with correctness. The paper discovers LRMs implicitly know when to stop thinking. SAGE (Self-Aware Guided Efficient Reasoning) exploits this via a novel sampling paradigm, improving accuracy and efficiency on mathematical benchmarks.

Summary generated by Claude — human-verified