KohakuRAG: A simple RAG framework with hierarchical document indexing

TL;DR

KohakuRAG introduces a hierarchical document indexing framework with improved retrieval, stability, and citation accuracy for question-answering systems, achieving top results on the WattBot 2025 Challenge.

Contribution

It presents a novel four-level tree document structure, an LLM-powered query planner, and ensemble inference techniques for enhanced RAG performance.

Findings

Achieved first place on WattBot 2025 Challenge leaderboard.

Hierarchical dense retrieval matches hybrid approaches in effectiveness.

Prompt ordering and ensemble voting significantly improve accuracy.

Abstract

Retrieval-augmented generation (RAG) systems that answer questions from document collections face compounding difficulties when high-precision citations are required: flat chunking strategies sacrifice document structure, single-query formulations miss relevant passages through vocabulary mismatch, and single-pass inference produces stochastic answers that vary in both content and citation selection. We present KohakuRAG, a hierarchical RAG framework that preserves document structure through a four-level tree representation (document $\rightarrow$ section $\rightarrow$ paragraph $\rightarrow$ sentence) with bottom-up embedding aggregation, improves retrieval coverage through an LLM-powered query planner with cross-query reranking, and stabilizes answers through ensemble inference with abstention-aware voting. We evaluate on the WattBot 2025 Challenge, a benchmark requiring systems to answer technical questions from 32 documents with $\pm$0.1% numeric tolerance and exact source attribution. KohakuRAG achieves first place on both public and private leaderboards (final score 0.861), as the only team to maintain the top position across both evaluation partitions. Ablation studies reveal that prompt ordering (+80% relative), retry mechanisms (+69%), and ensemble voting with blank filtering (+1.2pp) each contribute substantially, while hierarchical dense retrieval alone matches hybrid sparse-dense approaches (BM25 adds only +3.1pp). We release KohakuRAG as open-source software at https://github.com/KohakuBlueleaf/KohakuRAG.