Question Answering on HotpotQA (EM, Avg)
46.1EMMultiQuery + GeoRAG
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MultiQuery + GeoRAGCandidates=top-200→52026.06 | 46.1 | 56.4 | |
| GraphRAG + GeoRAGCandidates=top-200→52026.06 | 45.7 | 56.4 | |
| HyDE + GeoRAGCandidates=top-200→52026.06 | 44.9 | 55.4 | |
| Hybrid RRF + GeoRAGCandidates=top-200→52026.06 | 44.2 | 55.7 | |
| Dense + GeoRAGCandidates=top-200→52026.06 | 43.5 | 54.7 | |
| BM25 + GeoRAGCandidates=top-200→52026.06 | 40.8 | 53.2 | |
| GraphRAG (direct)Candidates=top-52026.06 | 37.8 | 49.7 | |
| MultiQuery (direct)Candidates=top-52026.06 | 37.2 | 49.5 | |
| HyDE (direct)Candidates=top-52026.06 | 36.1 | 48.8 | |
| Hybrid RRF (direct)Candidates=top-52026.06 | 35.4 | 49.2 | |
| Dense (direct)Candidates=top-52026.06 | 33.8 | 47.2 | |
| BM25 (direct)Candidates=top-52026.06 | 31.2 | 46 |