用慢思考机制减少大模型幻觉，提升生成可靠性
Think More, Hallucinate Less: Mitigating Hallucinations via Dual Process of Fast and Slow Thinking
- 引入树搜索与自评奖励模型，实现逐步推理生成
- 在多语言数据集上显著降低幻觉率，优于基线方法
- 基于双过程理论动态切换快慢思考，兼顾效率与准确
大语言模型虽能力卓越，但仍存在幻觉问题。传统自回归生成缺乏深思熟虑，常导致不可靠的回答。本文提出HaluSearch框架，通过树搜索算法（如MCTS）实现显式的慢思考生成过程，将文本生成视为逐步推理任务，利用自评价奖励模型对每一步生成进行打分，引导搜索走向最可靠的路径，充分挖掘模型内部知识。为平衡效率与质量，设计受认知科学双过程理论启发的分层思维切换机制，在实例和步骤层面动态切换快慢思考模式，适应问题复杂度与推理状态。在中英文数据集上的大量实验表明，该方法显著优于基线模型。
原文摘要 · Abstract (English)
Large language models (LLMs) demonstrate exceptional capabilities, yet still face the hallucination issue. Typical text generation approaches adopt an auto-regressive generation without deliberate reasoning, which often results in untrustworthy and factually inaccurate responses. In this paper, we propose HaluSearch, a novel framework that incorporates tree search-based algorithms (e.g. MCTS) to enable an explicit slow thinking generation process for mitigating hallucinations of LLMs during inference. Specifically, HaluSearch frames text generation as a step-by-step reasoning process, using a self-evaluation reward model to score each generation step and guide the tree search towards the most reliable generation pathway for fully exploiting the internal knowledge of LLMs. To balance efficiency and quality, we introduce a hierarchical thinking system switch mechanism inspired by the dual process theory in cognitive science, which dynamically alternates between fast and slow thinking modes at both the instance and step levels, adapting to the complexity of questions and reasoning states. We conduct extensive experiments on both English and Chinese 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。