| Benchmark | Model | Rank | Results |
|---|---|---|---|
| visual-commonsense-reasoning-on-gd-vcr | Human | #1 | Accuracy: 88.84 |
| visual-commonsense-reasoning-on-gd-vcr | ViLBERT | #2 | Accuracy: 59.99Gap (West): -7.28 |
| visual-commonsense-reasoning-on-gd-vcr | VisualBERT | #3 | Accuracy: 53.95Gap (West): -10.42 |
| visual-commonsense-reasoning-on-gd-vcr | Text-only BERT | #4 | Accuracy: 35.33 |