Multi-hop Reasoning via Early Knowledge Alignment
Yuxin Wang, Shicheng Fang, Bo Wang, Qi Luo, Xuanjing Huang, Yining Zheng, Xipeng Qiu
TL;DR
Early Knowledge Alignment (EKA) introduces a grounding step that aligns LLM reasoning with the retrieval corpus before planning in iterative RAG. By seeding the first Think with context from top-$k$ retrieved passages, EKA reduces entropy-driven exploration, mitigates cascading errors, and improves both retrieval precision and answer quality across multiple RL-based RAG settings. The approach is demonstrated to be effective in training, inference, and training-free scenarios, with robust generalization to diverse datasets and retrievers. These findings highlight the importance of grounding prior to iterative search and reasoning, and they offer a scalable, plug-and-play enhancement for large-scale RAG systems.
Abstract
Retrieval-Augmented Generation (RAG) has emerged as a powerful paradigm for Large Language Models (LLMs) to address knowledge-intensive queries requiring domain-specific or up-to-date information. To handle complex multi-hop questions that are challenging for single-step retrieval, iterative RAG approaches incorporating reinforcement learning have been proposed. However, existing iterative RAG systems typically plan to decompose questions without leveraging information about the available retrieval corpus, leading to inefficient retrieval and reasoning chains that cascade into suboptimal performance. In this paper, we introduce Early Knowledge Alignment (EKA), a simple but effective module that aligns LLMs with retrieval set before planning in iterative RAG systems with contextually relevant retrieved knowledge. Extensive experiments on six standard RAG datasets demonstrate that by establishing a stronger reasoning foundation, EKA significantly improves retrieval precision, reduces cascading errors, and enhances both performance and efficiency. Our analysis from an entropy perspective demonstrate that incorporating early knowledge reduces unnecessary exploration during the reasoning process, enabling the model to focus more effectively on relevant information subsets. Moreover, EKA proves effective as a versatile, training-free inference strategy that scales seamlessly to large models. Generalization tests across diverse datasets and retrieval corpora confirm the robustness of our approach. Overall, EKA advances the state-of-the-art in iterative RAG systems while illuminating the critical interplay between structured reasoning and efficient exploration in reinforcement learning-augmented frameworks. The code is released at \href{https://github.com/yxzwang/EarlyKnowledgeAlignment}{Github}.