让大模型分步思考该不该查资料，减少幻觉并提升答案准确率。
DeepRAG: Thinking to Retrieve Step by Step for Large Language Models
- 将检索增强推理建模为马尔可夫决策过程，分步判断是否需外部检索
- 在测试中使回答准确率提升26.4%，同时减少冗余检索
- 适合需要高可信度问答的场景，如医疗、法律等专业领域
大语言模型虽具备强大推理能力，但受限于参数化知识的时间性、准确性和全面性，常出现事实性幻觉。现有检索增强生成（RAG）方法因任务分解不当和重复检索，引入噪声并降低响应质量。本文提出DeepRAG框架，将检索增强推理建模为马尔可夫决策过程（MDP），实现合理且自适应的检索。通过迭代分解查询，模型在每一步动态决定是否检索外部知识或依赖参数化推理。实验表明，DeepRAG显著提升检索效率，使答案准确率提高26.4%，验证了其在增强检索-推理方面的有效性。
原文摘要 · Abstract (English)
Large Language Models (LLMs) have shown remarkable reasoning capabilities, while their practical applications are limited by severe factual hallucinations due to limitations in the timeliness, accuracy, and comprehensiveness of their parametric knowledge. Meanwhile, enhancing retrieval-augmented generation (RAG) with reasoning remains challenging due to ineffective task decomposition and redundant retrieval, which can introduce noise and degrade response quality. In this paper, we propose DeepRAG, a framework that models retrieval-augmented reasoning as a Markov Decision Process (MDP), enabling reasonable and adaptive retrieval. By iteratively decomposing queries, DeepRAG dynamically determines whether to retrieve external knowledge or rely on parametric reasoning at each step. Experiments show that DeepRAG improves retrieval efficiency and boosts answer accuracy by 26.4%, demonstrating its effectiveness in enhancing retrieval-augmented reasoning.
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。