Image Classification on CIFAR10 (ACC and CEA)
89.8AccuracyPS-CBM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| PS-CBMBackbone=CLIP RN502025.11 | 89.8 | 68.5 | |
| VLG-CBMBackbone=CLIP RN502025.11 | 89.6 | 67.4 | |
| Linear ProbeBackbone=CLIP RN502025.11 | 88.5 | — | |
| DN-CBMBackbone=CLIP RN502025.11 | 88.2 | 55.2 | |
| Res-CBMBackbone=CLIP RN502025.11 | 87.6 | 61.9 | |
| LaBoBackbone=CLIP RN502025.11 | 87.5 | 60.1 | |
| LF-CBMBackbone=CLIP RN502025.11 | 87.3 | 63.1 | |
| V2C-CBMBackbone=CLIP RN502025.11 | 87.3 | 60 | |
| DCBMBackbone=CLIP RN502025.11 | 85.7 | 55.9 | |
| LM4CVBackbone=CLIP RN502025.11 | 81.4 | 62.8 |