CompactRAG: Reducing LLM Calls and Token Overhead in Multi-Hop Question Answering

TL;DR

CompactRAG significantly reduces the number of LLM calls and token usage in multi-hop question answering by restructuring knowledge offline and carefully decomposing queries online, maintaining accuracy while improving efficiency.

Contribution

It introduces a decoupled framework that minimizes LLM interactions in multi-hop RAG, enabling efficient reasoning with fewer calls and lower token consumption.

Findings

Achieves competitive accuracy on multiple datasets.

Reduces token consumption compared to iterative RAG.

Invokes LLM only twice during inference regardless of reasoning hops.

Abstract

Retrieval-augmented generation (RAG) has become a key paradigm for knowledge-intensive question answering. However, existing multi-hop RAG systems remain inefficient, as they alternate between retrieval and reasoning at each step, resulting in repeated LLM calls, high token consumption, and unstable entity grounding across hops. We propose CompactRAG, a simple yet effective framework that decouples offline corpus restructuring from online reasoning. In the offline stage, an LLM reads the corpus once and converts it into an atomic QA knowledge base, which represents knowledge as minimal, fine-grained question-answer pairs. In the online stage, complex queries are decomposed and carefully rewritten to preserve entity consistency, and are resolved through dense retrieval followed by RoBERTa-based answer extraction. Notably, during inference, the LLM is invoked only twice in total - once for sub-question decomposition and once for final answer synthesis - regardless of the number of reasoning hops. Experiments on HotpotQA, 2WikiMultiHopQA, and MuSiQue demonstrate that CompactRAG achieves competitive accuracy while substantially reducing token consumption compared to iterative RAG baselines, highlighting a cost-efficient and practical approach to multi-hop reasoning over large knowledge corpora. The implementation is available at GitHub.