PAC Reasoning：为大语言模型高效推理控制性能损失
PAC Reasoning: Controlling the Performance Loss for Efficient Reasoning
AI 导读
一项被 NeurIPS 2026 接收为海报论文的研究提出 PAC Reasoning（Probably Approximately Correct Reasoning），通过构建性能损失的上置信界并设定切换阈值，让大推理模型在思考与非思考模式间动态切换。该方法以 distribution-free 方式保证性能损失不超过用户指定容忍度，在推理基准上验证了节省计算预算并控制性能损失的效果。
正文 · 原文
Abstract:Large reasoning models (LRMs) have achieved remarkable progress in complex problem-solving tasks. Despite this success, LRMs typically suffer from high computational costs during deployment, highlighting a need for efficient inference. A practical direction is to switch the LRM between thinking and non-thinking modes dynamically. However, such approaches often introduce additional reasoning errors and lack statistical guarantees for the performance loss, which are critical for high-stakes applications. In this work, we propose Probably Approximately Correct (PAC) Reasoning that controls the performance loss under the user-specified tolerance. Specifically, we construct an upper confidence bound on the performance loss and determine a threshold for switching to the non-thinking model. Theoretically, using the threshold to switch between the thinking and non-thinking modes ensures bounded performance loss in a distribution-free manner. Our experiments on reasoning benchmarks show that the proposed method can save computational budgets and control the user-specified performance loss.
| Comments: | Accepted at NeurIPS 2026 (poster). Camera-ready version |
| Subjects: | Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Statistics Theory (math.ST) |
| Cite as: | arXiv:2510.09133 [cs.AI] |
| (or arXiv:2510.09133v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2510.09133 arXiv-issued DOI via DataCite |
Submission history
From: Hao Zeng [view email]
[v1]
Fri, 10 Oct 2025 08:33:47 UTC (148 KB)
[v2]
Fri, 30 Jan 2026 09:59:29 UTC (498 KB)
[v3]
Wed, 11 Nov 2026 00:35:09 UTC (316 KB)
来源：arXiv cs.AI · arxiv.org