动态调整推理长度，让大模型该快则快、该慢则慢。
Adaptive Overclocking: Dynamic Control of Thinking Path Length via Real-Time Reasoning Signals
- 根据每步输出的不确定性实时调节推理速度
- 在GSM8K等数据集上降低复杂任务延迟30%以上
- 适合需要高效推理的智能问答系统
大型推理模型常因过度思考导致计算效率低下，固定推理预算难以匹配任务复杂度差异。本文提出自适应超频（Adaptive Overclocking），使超频参数α动态且上下文感知。通过两种互补信号实现：(1) 基于词元级别的模型不确定性，实现细粒度步骤控制；(2) 输入复杂度估计，用于初始化。我们设计了三种策略：不确定性感知α调度（UA-αS）、复杂度引导α初始化（CG-αI）及融合两者的优势自适应控制（HAC）。在GSM8K、MATH和SVAMP上的实验表明，HAC在准确率与延迟之间取得更优平衡，在简单问题上减少不必要的计算，同时为难题分配更多资源。该方法有效缓解过度思考，提升整体推理效率与性能。
原文摘要 · Abstract (English)
Large Reasoning Models (LRMs) often suffer from computational inefficiency due to overthinking, where a fixed reasoning budget fails to match the varying complexity of tasks. To address this issue, we propose Adaptive Overclocking, a method that makes the overclocking hyperparameter $α$ dynamic and context-aware. Our method adjusts reasoning speed in real time through two complementary signals: (1) token-level model uncertainty for fine-grained step-wise control, and (2) input complexity estimation for informed initialization. We implement this approach with three strategies: Uncertainty-Aware Alpha Scheduling (UA-$α$S), Complexity-Guided Alpha Initialization (CG-$α$I), and a Hybrid Adaptive Control (HAC) that combines both. Experiments on GSM8K, MATH, and SVAMP show that HAC achieves superior accuracy-latency trade-offs, reducing unnecessary computation on simple problems while allocating more resources to challenging ones. By mitigating overthinking, Adaptive Overclocking enhances both efficiency and overall reasoning performance.
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。