用自反思树搜索让大模型生成有据可查的文本，减少幻觉。
Think&Cite: Improving Attributed Text Generation with Self-Guided Tree Search and Progress Reward Modeling
- 通过自引导蒙特卡洛树搜索，让模型逐步推理并选择更优生成路径。
- 在三个数据集上，生成内容的准确率和引用匹配度均显著提升。
- 适合需要高可信度文本生成的应用，如科研写作、新闻报道。
尽管大语言模型表现出色，但容易产生幻觉和事实错误。为此，研究者提出有引证的文本生成方法，要求模型生成内容时附带支持证据。本文提出新框架 Think&Cite，将有引证生成建模为多步推理与搜索结合的问题。提出自引导蒙特卡洛树搜索（SG-MCTS），利用大模型的自我反思能力来指导树的扩展过程。引入进展奖励建模（Progress Reward Modeling），从生成和引证两个方面评估从根节点到当前状态的进展。在三个数据集上的实验表明，该方法显著优于基线方法。
原文摘要 · Abstract (English)
Despite their outstanding capabilities, large language models (LLMs) are prone to hallucination and producing factually incorrect information. This challenge has spurred efforts in attributed text generation, which prompts LLMs to generate content with supporting evidence. In this paper, we propose a novel framework, called Think&Cite, and formulate attributed text generation as a multi-step reasoning problem integrated with search. Specifically, we propose Self-Guided Monte Carlo Tree Search (SG-MCTS), which capitalizes on the self-reflection capability of LLMs to reason about the intermediate states of MCTS for guiding the tree expansion process. To provide reliable and comprehensive feedback, we introduce Progress Reward Modeling to measure the progress of tree search from the root to the current state from two aspects, i.e., generation and attribution progress. We conduct extensive experiments on three datasets and the results show that our approach significantly outperforms baseline approaches.
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。