Multi-hop Question Answering on 2WikiMHQA 500 samples (test)
91.1F1 ScoreLLM-Wiki
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLM-Wiki2026.05 | 91.1 | 85.4 | |
| LightRAG2026.05 | 84.7 | 76.4 | |
| HippoRAG 22026.05 | 83.1 | 70.6 | |
| Vanilla RAGretrieval=Dense2026.05 | 81.5 | 72.4 | |
| Vanilla RAGretrieval=BM252026.05 | 79 | 68.4 | |
| GraphRAG2026.05 | 72 | 64.8 | |
| RAPTOR2026.05 | 70.7 | 65.2 | |
| Nonemode=Closed-book2026.05 | 63.8 | 54.6 |