用蒙特卡洛树搜索优化多跳问答路径，减少大模型幻觉。
PathFinder: MCTS and LLM Feedback-based Path Selection for Multi-Hop Question Answering
- 结合蒙特卡洛树搜索生成推理路径，提升路径多样性。
- 通过子答案召回和大模型判别过滤错误路径，提升训练数据质量。
- 自动重构失败检索的子问题，适合需要严谨推理的应用场景。
多跳问答是语言模型需经过多步推理才能得出正确答案的挑战性任务。尽管大型语言模型具备推理能力，现有基于训练的方法仍受大模型幻觉和错误推理路径影响。为此，我们提出 PATHFINDER：(i) 使用蒙特卡洛树搜索生成训练路径轨迹；(ii) 通过子答案召回和大模型作为裁判的验证机制，过滤错误与冗长轨迹，提升训练数据质量；(iii) 重构失败检索的子问题以增强鲁棒性。实验表明，该方法在多个公开基准数据集上显著提升了多跳问答性能。
原文摘要 · Abstract (English)
Multi-hop question answering is a challenging task in which language models must reason over multiple steps to reach the correct answer. With the help of Large Language Models and their reasoning capabilities, existing systems are able to think and decompose an input question over multiple steps to analyze, retrieve, and reason. However, training-based approaches for this problem still suffer from LLM hallucinations and incorrect reasoning paths that hinder performance. Hence, we propose PATHFINDER, an approach that: (i) uses Monte Carlo Tree Search to generate training path traces, (ii) improves training data quality by filtering erroneous and lengthy traces using sub-answer recall and LLM-as-a-judge verification, and (iii) reformulates sub-queries to handle failed retrieval cases. By following these steps, we demonstrate that PATHFINDER improves the performance of multi-hop QA over public benchmark datasets.
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。