DeepRAG: Thinking to Retrieve Step by Step for Large Language Models

2025/02/03 by Xinyan Guan, Guan, Xinyan, Jiali Zeng +15 · 9 voices · 4 citations
Computer Science · #Topic Modeling #Natural Language Processing Techniques

paper · pdf · doi:10.48550/arxiv.2502.01142

Abstract

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.

Citations

• Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning
• Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity

Cited by

• AdaSearch: Balancing Parametric Knowledge and Search in Large Language Models via Reinforcement Learning
• PathFinder: MCTS and LLM Feedback-based Path Selection for Multi-Hop Question Answering
• Improving LLM Reasoning via Dependency-Aware Query Decomposition and Logic-Parallel Content Expansion
• EviNote-RAG: Enhancing RAG Models via Answer-Supportive Evidence Notes

Discussions

• DeepRAG: Thinking to retrieval step by step for large language models [hn, 191 points, 29 comments]
• They implemented yaprag: arxiv.org/abs/2502.01142 [bsky, 1 points, 0 comments]
• DeepRAG: Thinking to Retrieval Step by Step for Large Language Models [bsky, 0 points, 0 comments]
• DeepRAG: Thinking to retrieval step by step for large language models https://arxiv.org/abs/2502.01142 https://news.ycombinator.com/item?id=42932948 [bsky, 0 points, 0 comments]
• DeepRAG: Thinking to Retrieval Step by Step for Large Language Models https://arxiv.org/abs/2502.01142 [bsky, 0 points, 0 comments]
• DeepRAG: Thinking to Retrieval Step by Step for Large Language Models https://arxiv.org/abs/2502.01142 (https://news.ycombinator.com/item?id=42932948) [bsky, 0 points, 0 comments]
• DeepRAG: Thinking to Retrieval Step by Step for Large Language Models (arxiv.org) Main Link | Discussion [bsky, 0 points, 0 comments]
• DeepRAG: Thinking to Retrieval Step by Step for Large Language Models https://arxiv.org/abs/2502.01142 [comments] [22 points] [bsky, 0 points, 0 comments]
• DeepRAG: Thinking to Retrieval Step by Step for Large Language Models https://arxiv.org/abs/2502.01142 (https://news.ycombinator.com/item?id=42932948) [bsky, 0 points, 0 comments]

Related

• Open-RAG: Enhanced Retrieval-Augmented Reasoning with Open-Source Large Language Models
• AssistRAG: Boosting the Potential of Large Language Models with an Intelligent Information Assistant
• Auto-RAG: Autonomous Retrieval-Augmented Generation for Large Language Models
• Rethinking with Retrieval: Faithful Large Language Model Inference
• Synergizing RAG and Reasoning: A Systematic Review