发现大模型推理冗余，提出动态终止机制节省44%计算量
The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis
- 通过分析推理长度与语义演化，识别出无需继续思考的临界点
- 在多个基准上实现最高44%的令牌消耗减少，准确率保持不变
- 适合需要高效推理的部署场景，尤其对长文本生成有帮助
通过显式推理轨迹进行测试时扩展能显著提升大语言模型性能，但常引发过度思考。我们从两个角度分析推理过程：推理长度动态揭示了思考时长与答案内容长度之间的补偿性权衡，最终导致推理冗余；推理语义动态则识别出语义收敛和重复振荡现象。这些动态揭示了实例特定的推理完成点（RCP），超出该点后计算不再带来性能提升。由于RCP因实例而异，我们提出推理完成点检测器（RCPD），一种基于监测终止标记（如</think>）秩动态的推理时早退方法。在AIME和GPQA基准上，使用Qwen3和DeepSeek-R1模型，RCPD将令牌使用量最多降低44%，同时保持准确率，为高效的测试时扩展提供了原则性方案。
原文摘要 · Abstract (English)
Test-time scaling via explicit reasoning trajectories significantly boosts large language model (LLM) performance but often triggers overthinking. To explore this, we analyze reasoning through two lenses: Reasoning Length Dynamics, which reveals a compensatory trade-off between thinking and answer content length that eventually leads to thinking redundancy, and Reasoning Semantic Dynamics, which identifies semantic convergence and repetitive oscillations. These dynamics uncover an instance-specific Reasoning Completion Point (RCP), beyond which computation continues without further performance gain. Since the RCP varies across instances, we propose a Reasoning Completion Point Detector (RCPD), an inference-time early-exit method that identifies the RCP by monitoring the rank dynamics of termination tokens (e.g., </think>). Across AIME and GPQA benchmarks using Qwen3 and DeepSeek-R1, RCPD reduces token usage by up to 44% while preserving accuracy, offering a principled approach to efficient test-time scaling.
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。