测试大模型能否发现长思维链中的错误，揭示现有模型的局限性。
Can Large Language Models Detect Errors in Long Chain-of-Thought Reasoning?
- 构建DeltaBench数据集，包含多个o1类模型生成的长思维链。
- 发现现有模型在检测长思维链错误时准确率不足50%。
- 适合研究推理能力、模型批判性评估的研究者参考。
近期，o1类模型因其生成长思维链（CoT）以提升大语言模型推理能力而受到广泛关注。本文引入DeltaBench数据集，包含来自QwQ、DeepSeek-R1等o1类模型在数学、编程和通用推理任务中生成的长思维链，旨在衡量现有大模型对长思维链中错误的检测能力。基于该数据集，我们首先对不同o1类模型生成的长思维链进行细粒度分析，评估其有效性和效率；随后对现有过程奖励模型（PRMs）和批判模型展开广泛评测，检测标注过程中的错误，探究现有模型在边界与局限性方面的表现。最终，期望DeltaBench能帮助开发者更深入理解自身模型在长思维链推理方面的能力。
原文摘要 · Abstract (English)
Recently, o1-like models have drawn significant attention, where these models produce the long Chain-of-Thought (CoT) reasoning steps to improve the reasoning abilities of existing Large Language Models (LLMs). In this paper, to understand the qualities of these long CoTs and measure the critique abilities of existing LLMs on these long CoTs, we introduce the DeltaBench, including the generated long CoTs from different o1-like models (e.g., QwQ, DeepSeek-R1) for different reasoning tasks (e.g., Math, Code, General Reasoning), to measure the ability to detect errors in long CoT reasoning. Based on DeltaBench, we first perform fine-grained analysis of the generated long CoTs to discover the effectiveness and efficiency of different o1-like models. Then, we conduct extensive evaluations of existing process reward models (PRMs) and critic models to detect the errors of each annotated process, which aims to investigate the boundaries and limitations of existing PRMs and critic models. Finally, we hope that DeltaBench could guide developers to better understand the long CoT reasoning abilities of their models.
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。