思维草图：基于自适应认知启发式草图的LLM高效推理
Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching
AI总结：
提出SoT提示框架，融合认知范式与语言约束，通过三种动态选择的推理模式在18个数据集上减少高达84%的token，同时保持甚至提升推理准确率。
摘要：
近年来，大语言模型（LLM）的进展通过思维链（CoT）提示实现了强大的推理能力，该提示引导逐步问题解决，但往往以中间输出过度冗长为代价，导致计算开销增加。我们提出思维草图（SoT），一个提示框架，将认知启发的推理范式与语言约束相结合，以减少token使用量同时保持推理准确性。SoT设计为灵活、模块化的方法，并实例化为三种范式——概念链、分块符号主义和专家词汇——每种针对不同推理任务，并通过轻量级路由模型在测试时动态选择。在跨越多个领域、语言和模态的18个推理数据集上，SoT实现了高达84%的token减少，且准确率损失极小。在数学和多跳推理等任务中，它甚至在缩短输出的同时提高了准确率。
英文摘要：
Recent advances in large language models (LLMs) have enabled strong reasoning capabilities through Chain-of-Thought (CoT) prompting, which elicits step-by-step problem solving, but often at the cost of excessive verbosity in intermediate outputs, leading to increased computational overhead. We propose Sketch-of-Thought (SoT), a prompting framework that integrates cognitively inspired reasoning paradigms with linguistic constraints to reduce token usage while preserving reasoning accuracy. SoT is designed as a flexible, modular approach and is instantiated with three paradigms--Conceptual Chaining, Chunked Symbolism, and Expert Lexicons--each tailored to distinct reasoning tasks and selected dynamically at test-time by a lightweight routing model. Across 18 reasoning datasets spanning multiple domains, languages, and modalities, SoT achieves token reductions of up to 84% with minimal accuracy loss. In tasks such as mathematical and multi-hop reasoning, it even improves accuracy while shortening outputs.
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。
