Multi-hop Reasoning on OF-General
37.1Accuracy (Correct)HEAR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| HEARCategory=HEAR, Hyperedges=Complete2026.05 | 37.1 | 88.7 | 7.3 | 81,600 | |
| HEARCategory=HEAR, Hyperedges=Declarative2026.05 | 35.8 | 86.2 | 7.4 | 76,700 | |
| Table-RAGCategory=Internal Baseline, Hyperedges=No2026.05 | 29.6 | 85.5 | 6.7 | 40,800 | |
| Table listCategory=Internal Baseline, Hyperedges=No2026.05 | 28.9 | 88.7 | 4.5 | 35,100 | |
| CSR-RAGCategory=External Baseline, Hyperedges=—2026.05 | 6.3 | 72.2 | — | 8,100 |