Multi-hop Question Answering on HotpotQA 500 samples (test)
83.9F1 ScoreLLM-Wiki
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLM-Wiki2026.05 | 83.9 | 71 | |
| LightRAG2026.05 | 81.9 | 68.2 | |
| HippoRAG 22026.05 | 80.5 | 66.8 | |
| RAPTOR2026.05 | 80.1 | 67.4 | |
| GraphRAG2026.05 | 77.1 | 65 | |
| Vanilla RAGretrieval=Dense2026.05 | 76.4 | 64.2 | |
| Vanilla RAGretrieval=BM252026.05 | 71.7 | 59 | |
| Nonemode=Closed-book2026.05 | 55.1 | 44.2 |