What AI says about LevelRAG

LevelRAG is a two-stage retrieval-augmented generation framework that enhances answer completeness and accuracy through multi-hop logic planning and hybrid retrieval. It first uses a high-level searcher to decompose a user query into atomic sub-queries, then employs multiple low-level searchers and LLMs to refine these queries and retrieve relevant documents for final answer generation. The repository provides the source code and installation guidance for LevelRAG, built on the FlexRAG project to reproduce the approach described in the LevelRAG paper.