让大模型自动判断何时停止思考，节省计算资源。
Thought calibration: Efficient and confident test-time scaling
- 根据推理过程动态决定是否继续思考
- 在分布内数据上减少60%的思考令牌
- 适合追求高效推理的部署场景
推理型大语言模型通过延长思考时间实现性能提升，但代价是巨大的计算开销。直接限制测试时预算会损害整体表现，但并非所有问题难度相同。本文提出思想校准（thought calibration），动态决定何时终止思考。将语言模型不断积累的思想视为嵌套的推理树结构，目标是识别出新推理趋于停滞的节点。通过轻量级探测器作用于模型隐藏表示，有效捕捉推理结构与答案一致性信息。在三个推理模型和四个数据集上验证，该方法在分布内数据上可将思考令牌减少高达60%，分布外数据上减少20%，同时保持原有性能。
原文摘要 · Abstract (English)
Reasoning large language models achieve impressive test-time scaling by thinking for longer, but this performance gain comes at significant compute cost. Directly limiting test-time budget hurts overall performance, but not all problems are equally difficult. We propose thought calibration to decide dynamically when thinking can be terminated. To calibrate our decision rule, we view a language model's growing body of thoughts as a nested sequence of reasoning trees, where the goal is to identify the point at which novel reasoning plateaus. We realize this framework through lightweight probes that operate on top of the language model's hidden representations, which are informative of both the reasoning structure and overall consistency of response. Based on three reasoning language models and four datasets, thought calibration preserves model performance with up to a 60% reduction in thinking tokens on in-distribution data, and up to 20% in out-of-distribution data.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。