Multi-hop Question Answering on SAGE-generated In-domain (test)
803-Hop Accuracy
gemini-2.5-flash
Updated 3d ago
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
2026.01 | 80 | 67 | 57.3 | 48.7 | 37.3 | 58.1 | |
2026.01 | 68.7 | 55.7 | 50.3 | 43.3 | 41.3 | 51.9 | |
2026.01 | 55.7 | 38 | 35.7 | 37.3 | 24 | 38.1 | |
2026.01 | 48.3 | 28.7 | 24.7 | 25 | 21.2 | 29.6 | |
2026.01 | 45 | 26.3 | 25.7 | 24.7 | 23.6 | 29.1 | |
2026.01 | 42.3 | 26.7 | 25.3 | 25 | 23.3 | 28.5 | |
2026.01 | 37.3 | 20 | 18.3 | 19 | 17.1 | 22.4 | |
2026.01 | 25.3 | 12 | 15.3 | 11.7 | 15 | 15.9 |