思维的演变：通过推理动态分析追踪LLM过度思考
The Evolution of Thought: Tracking LLM Overthinking via Reasoning Dynamics Analysis
AI总结：
通过分析推理动态，提出RCPD方法以识别LLM推理完成点，减少token使用量同时保持准确性。
摘要：
通过显式推理轨迹进行测试时扩展显著提升了大语言模型（LLM）的性能，但往往引发过度思考。为了探讨这一现象，我们通过两个视角分析推理：推理长度动态，揭示了思考与答案内容长度之间的补偿性权衡，最终导致思考冗余；推理语义动态，识别了语义收敛和重复振荡。这些动态揭示了实例特定的推理完成点（RCP），在此之后计算继续进行而不再带来性能提升。由于RCP在不同实例中变化，我们提出了推理完成点检测器（RCPD），一种推理时的早退方法，通过监控终止标记（例如，）的排名动态来识别RCP。在使用Qwen3和DeepSeek-R1进行AIME和GPQA基准测试时，RCPD将token使用量减少高达44%，同时保持准确性，提供了一种系统的方法来高效地进行测试时扩展。
英文摘要：
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.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
