让大模型的推理过程可验证，提升答案可信度
VeriCoT: Neuro-symbolic Chain-of-Thought Validation via Logical Consistency Checks
- 将思维链转为一阶逻辑形式，自动检测推理漏洞
- 在三个数据集上验证错误推理准确率超90%
- 适合需要高可靠性的法律、医疗等专业场景
大模型通过思维链（CoT）进行多步推理，但无法自证逻辑正确性。即使答案正确，推理过程也可能存在漏洞，影响高风险场景的信任度。为此，我们提出VeriCoT，一种神经符号方法，将CoT每一步转化为一阶逻辑，并识别支撑论证的来源：原始上下文、常识知识或先前推理步骤。符号化表示使自动求解器可验证逻辑有效性，自然语言前提则便于人类与系统发现无根据或谬误推理。在ProofWriter、LegalBench和BioASQ数据集上的实验表明，VeriCoT能有效识别错误推理，并作为最终答案正确性的强预测指标。我们进一步利用其验证信号实现：(1) 推理时自我反思，(2) 在VeriCoT提炼的数据集上进行监督微调（SFT），(3) 基于验证信号的成对奖励进行偏好微调（PFT）与直接偏好优化（DPO），显著提升推理有效性和准确性。
原文摘要 · Abstract (English)
LLMs can perform multi-step reasoning through Chain-of-Thought (CoT), but they cannot reliably verify their own logic. Even when they reach correct answers, the underlying reasoning may be flawed, undermining trust in high-stakes scenarios. To mitigate this issue, we introduce VeriCoT, a neuro-symbolic method that extracts and verifies formal logical arguments from CoT reasoning. VeriCoT formalizes each CoT reasoning step into first-order logic and identifies premises that ground the argument in source context, commonsense knowledge, or prior reasoning steps. The symbolic representation enables automated solvers to verify logical validity while the NL premises allow humans and systems to identify ungrounded or fallacious reasoning steps. Experiments on the ProofWriter, LegalBench, and BioASQ datasets show VeriCoT effectively identifies flawed reasoning, and serves as a strong predictor of final answer correctness. We also leverage VeriCoT's verification signal for (1) inference-time self-reflection, (2) supervised fine-tuning (SFT) on VeriCoT-distilled datasets and (3) preference fine-tuning (PFT) with direct preference optimization (DPO) using verification-based pairwise rewards, further improving reasoning validity and accuracy.
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。