Deductive Verification of Chain-of-Thought Reasoning
Zhan Ling1 Yunhao Fang Xuanlin Li1 Zhiao Huang1 Mingu Lee2 Roland Memisevic2 Hao Su1 1UC San Diego, 2Qualcomm AI Research Equal contributionQualcomm AI Research is an initiative of Qualcomm Technologies, Inc
Abstract
Large Language Models (LLMs) significantly benefit from Chain-of-Thought (CoT) prompting in performing various reasoning tasks. While CoT allows models to produce more comprehensive reasoning processes, its emphasis on intermediate reasoning steps can inadvertently introduce hallucinations and accumulated errors, thereby limiting models’ ability to solve complex reasoning tasks. Inspired by how humans engage in careful and meticulous deductive logical reasoning processes to solve tasks, we seek to enable language models to perform explicit and rigorous deductive reasoning, and also ensure the trustworthiness of their reasoning process through self-verification. However, directly verifying the validity of an entire deductive reasoning process is challenging, even with advanced models like ChatGPT. In light of this, we propose to decompose a reasoning verification process into a series of step-by-step subprocesses, each only receiving their necessary context and premises. To facilitate this procedure, we propose Natural Program, a natural language-based deductive reasoning format. Our approach enables models to generate precise reasoning steps where subsequent steps are more rigorous
中文速览
让大语言模型（LLM）推理更可靠是当前的核心难题——哪怕链式思维（Chain-of-Thought, CoT）提示能让模型写出更长的推理过程，中间步骤仍然频繁出现幻觉和累积错误。为此，研究者提出了一种名为"自然程序"（Natural Program）的自然语言演绎推理格式，让模型像写形式化证明一样逐步列出前提与结论，并把对整条推理链的验证拆解为一系列只需要少量必要前提的局部子验证任务，从而让 GPT-3.5-turbo 等现有模型无需微调、仅凭上下文学习就能逐步自我核验每一个推理步骤的有效性。实验表明，通过将这种逐步验证与采样投票相结合，模型在算术和常识推理等多个基准数据集上不仅答案正确率得到提升，更重要的是推理过程的严谨性和可信度显著增强——当反复采样仍无法得到通过验证的推理链时，模型还会输出"未知"以避免误导用户。这项工作的价值在于，它将目标从单纯追求答案正确转向追求推理过程本身的可靠性，为构建值得信赖的 AI 推理系统提供了一条无需大规模重训练的实用路径。
原文 arXiv:2306.03872;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2306.03872v3