| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Compositional Zero-Shot Learning | MIT-States open world | HM37.2 | 38 | |
| Compositional Zero-Shot Learning | MIT-States Closed World | Harmonic Mean (HM)0.413 | 32 | |
| Compositional Zero-Shot Learning | MIT-States Closed World (test) | AUC23.4 | 12 | |
| Classification | MIT-states | Top-1 Acc52.5 | 12 | |
| Generalized Compositional Zero-Shot Learning | MIT-States (test) | AUC (1)6.5 | 12 | |
| Compositional Zero-Shot Learning | MIT-States (test) | Top-1 Acc19.9 | 11 | |
| Composed Image Retrieval | MIT-States (val) | R@113.9 | 9 | |
| Unseen Pair Detection | MIT-States (test) | Accuracy (Closed Set)14.8 | 8 | |
| Compositional Zero-Shot Learning | MIT-States 11 (test) | Seen Accuracy38.6 | 8 | |
| Text-based Image Retrieval | MIT-States | Rank-1 Accuracy15 | 7 | |
| Generalized Compositional Zero-Shot Learning | MIT-States (val) | AUC (1)4.3 | 6 | |
| Unseen Combination Classification | MIT-States (compositional) | Accuracy15.2 | 6 | |
| Image Classification | MIT-States Filtered to 97 classes (test) | Precision63.5 | 5 | |
| Image Classification | MIT-states | Top-1 Acc45.5 | 4 | |
| Compositional Zero-Shot Learning | MIT-States | Seen Accuracy (S)- | 0 | |
| Unseen Pair Detection | MIT-States | Accuracy (closed)- | 0 |