用分层推理链提升大模型的逻辑能力与效率
Hierarchical Chain-of-Thought: Enhancing LLM Reasoning Performance and Efficiency
- 将推理过程拆分为规划与执行的层级结构
- 平均准确率提升6.2%，推理长度减少13.9%
- 适合需要复杂逻辑推演的数学与认知任务
Chain-of-Thought (CoT) 提示显著提升了大语言模型（LLMs）的推理能力。然而，传统 CoT 常依赖无结构的扁平推理链，存在冗余和性能不佳的问题。本文提出分层推理链（Hi-CoT），一种专为复杂多步推理设计的结构化推理范式。Hi-CoT 通过交替进行指令规划与逐步执行，将推理过程分解为层级子步骤，使模型更有效地处理长距离推理并保持逻辑连贯性。在多种 LLM 和数学推理基准上的广泛评估显示，与 CoT 相比，Hi-CoT 平均准确率提升 6.2%（某些模型与任务达 61.4%），同时推理轨迹长度减少 13.9%。进一步研究表明，严格遵循层级结构可最大化准确率与效率。代码已公开于 https://github.com/XingshuaiHuang/Hi-CoT。
原文摘要 · Abstract (English)
Chain-of-Thought (CoT) prompting has significantly improved the reasoning capabilities of large language models (LLMs). However, conventional CoT often relies on unstructured, flat reasoning chains that suffer from redundancy and suboptimal performance. In this work, we introduce Hierarchical Chain-of-Thought (Hi-CoT), a structured reasoning paradigm specifically designed to address the challenges of complex, multi-step reasoning. Hi-CoT decomposes the reasoning process into hierarchical substeps by alternating between instructional planning and step-by-step execution. This decomposition enables LLMs to better manage long reasoning horizons and maintain logical coherence. Extensive evaluations across diverse LLMs and mathematical reasoning benchmarks show that Hi-CoT consistently improves average accuracy by 6.2% (up to 61.4% on certain models and tasks) while reducing reasoning trace length by 13.9% compared to CoT. We further show that accuracy and efficiency are maximized when models strictly adhere to the hierarchical structure. Our code is available at https://github.com/XingshuaiHuang/Hi-CoT.
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。