TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning

TL;DR

This paper introduces Terminator, a strategy for early stopping in large reasoning models that reduces reasoning length and inference time without sacrificing accuracy by predicting optimal exit points.

Contribution

The paper proposes a novel early-exit method trained on a new dataset to effectively determine when to stop reasoning in large models, reducing overthinking and computation.

Findings

Achieves 14%-55% reduction in reasoning length across four datasets.

Reduces inference latency by more than 2x compared to baseline models.

Outperforms current state-of-the-art early stopping methods.

Abstract

Large Reasoning Models (LRMs) achieve impressive performance on complex reasoning tasks via Chain-of-Thought (CoT) reasoning, which enables them to generate intermediate thinking tokens before arriving at the final answer. However, LRMs often suffer from significant overthinking, spending excessive compute time even after the answer is generated early on. Prior work has identified the existence of an optimal reasoning length such that truncating reasoning at this point significantly shortens CoT outputs with virtually no change in performance. However, determining optimal CoT lengths for practical datasets is highly non-trivial as they are fully task and model-dependent. In this paper, we precisely address this and design Terminator, an early-exit strategy for LRMs at inference to mitigate overthinking. The central idea underpinning Terminator is that the first arrival of an LRM's final answer is often predictable, and we leverage these first answer positions to create a novel dataset of optimal reasoning lengths to train Terminator. Powered by this approach, Terminator achieves significant reductions in CoT lengths of 14%-55% on average across four challenging practical datasets: MATH-500, AIME 2025, HumanEval, and GPQA, while outperforming current state-of-the-art methods and reducing inference latency by more than 2x compared to the original LRM.