Knowledge Base Question Answering on GrailQA 500-sample (dev)
84.7F1 ScoreRetinaQA
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RetinaQA#Train samples=44,337 (100%)2023.11 | 84.7 | 78.4 | |
| FuSIC-KBQA(T,P)#Train samples=44,337 (100%)2023.11 | 83.6 | 75 | |
| Pangu#Train samples=44,337 (100%)2023.11 | 83.5 | 76.2 | |
| TIARA#Train samples=44,337 (100%)2023.11 | 82.6 | 76 | |
| FuSIC-KBQA(P)#Train samples=44,337 (100%)2023.11 | 82.5 | 72.8 | |
| FuSIC-KBQA(T)#Train samples=44,337 (100%)2023.11 | 79.1 | 70.2 | |
| FuSIC-KBQA(T,P)#Train samples=4,434 (10%)2023.11 | 75.6 | 65.2 | |
| FuSIC-KBQA(T)#Train samples=4,434 (10%)2023.11 | 74.8 | 63 | |
| FuSIC-KBQA(T,P)#Train samples=2,217 (5%)2023.11 | 74.5 | 63.8 | |
| FuSIC-KBQA(P)#Train samples=4,434 (10%)2023.11 | 74.1 | 63 | |
| FuSIC-KBQA(P)#Train samples=2,217 (5%)2023.11 | 72.9 | 62 | |
| Pangu#Train samples=4,434 (10%)2023.11 | 72.4 | 64.6 | |
| FuSIC-KBQA(T)#Train samples=2,217 (5%)2023.11 | 70.6 | 60.4 | |
| Pangu#Train samples=2,217 (5%)2023.11 | 69.5 | 62.8 | |
| RetinaQA#Train samples=4,434 (10%)2023.11 | 68.3 | 62.4 | |
| TIARA#Train samples=4,434 (10%)2023.11 | 67.2 | 61.2 | |
| TIARA#Train samples=2,217 (5%)2023.11 | 64.8 | 58.2 | |
| RetinaQA#Train samples=2,217 (5%)2023.11 | 62.4 | 57 |