Research
Settle: Learning When to Stop Reasoning
arXiv:2609.38997v1 Announce Type: new Abstract: Reasoning models often continue generating after their answers have settled. Settle learns when to stop from answer stability in completed traces. It tr
arXiv:2609.38997v1 Announce Type: new Abstract: Reasoning models often continue generating after their answers have settled. Settle learns when to stop from answer stability in completed traces. It trains the existing end-of-reasoning token while keeping other predictions close to the base model, and requires only ordinary decoding at inference. On MATH-500 with Qwen3-4B, Settle reduces token count by 40% with a 0.5-percentage-point decrease in accuracy. It gains 6.16 percentage points over supervised fine-tuning on the same traces shortened at their first stable answer, at nearly identical token counts. Its stopping score predicts whether a correct answer will remain correct. Settle extends the accuracy-token-count Pareto frontier of the evaluated stopping methods.
Related
- Know When to Stop: Segment-Level Credit Assignment for Reducing Overthinking
- BLADE: Boundary-Expanded and Layer-Adaptive Dynamic Exit for Efficient LLM Reasoning
- Large Language Models Decide Early and Explain Later
Source: arXiv cs.CL | 2026-10-01