推理法庭：融合推理、行动与判断的多跳推理
Reasoning Court: Combining Reasoning, Action, and Judgment for Multi-Hop Reasoning
AI总结：
针对LLMs在多跳推理中存在幻觉和错误传播问题，提出推理法庭框架，引入LLM裁判评估多智能体生成的候选答案及推理，在无任务微调下性能优于现有少样本方法。
摘要：
尽管大型语言模型（LLMs）在问答和事实核查等任务中展现出强大能力，但仍存在幻觉和推理错误问题，尤其在需要整合多信息源的多跳任务中。现有方法通过基于检索的技术（将推理锚定在外部证据）、基于推理的方法（通过改进提示增强连贯性）或结合两者的混合策略解决这些问题。ReAct作为一种突出的混合方法，性能优于纯检索或纯推理方法；但它缺乏对中间推理步骤的内部验证，导致潜在错误在复杂推理任务中传播。本文提出推理法庭（Reasoning Court，RC）框架，扩展ReAct等迭代推理-检索方法，引入专用LLM裁判。与ReAct不同，RC利用该裁判独立评估由不同LLM智能体生成的多个候选答案及相关推理。裁判需基于呈现的推理和证据选择最具事实依据且逻辑连贯的答案，若所有候选答案不足、有缺陷或无效，则利用可用证据和预训练知识合成新答案。在多跳基准（HotpotQA、MuSiQue）和事实核查（FEVER）上的评估表明，RC在无任务特定微调的情况下持续优于最先进的少样本提示方法。
英文摘要：
While large language models (LLMs) have demonstrated strong capabilities in tasks like question answering and fact verification, they continue to suffer from hallucinations and reasoning errors, especially in multi-hop tasks that require integration of multiple information sources. Current methods address these issues through retrieval-based techniques (grounding reasoning in external evidence), reasoning-based approaches (enhancing coherence via improved prompting), or hybrid strategies combining both elements. One prominent hybrid method, ReAct, has outperformed purely retrieval-based or reasoning-based approaches; however, it lacks internal verification of intermediate reasoning steps, allowing potential errors to propagate through complex reasoning tasks. In this paper, we introduce Reasoning Court (RC), a novel framework that extends iterative reasoning-and-retrieval methods, such as ReAct, with a dedicated LLM judge. Unlike ReAct, RC employs this judge to independently evaluate multiple candidate answers and their associated reasoning generated by separate LLM agents. The judge is asked to select the answer that it considers the most factually grounded and logically coherent based on the presented reasoning and evidence, or synthesizes a new answer using available evidence and its pre-trained knowledge if all candidates are inadequate, flawed, or invalid. Evaluations on multi-hop benchmarks (HotpotQA, MuSiQue) and fact-verification (FEVER) demonstrate that RC consistently outperforms state-of-the-art few-shot prompting methods without task-specific fine-tuning.
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。
