Class-Incremental Learning on ImageNet-R 10 steps (incremental)
83.48Average AccuracyBoosting-CL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Boosting-CLBackbone=CLIP ViT-B/162026.03 | 83.48 | 74.81 | |
| SimEBackbone=CLIP ViT-B/162026.03 | 83.19 | 75.82 | |
| MISABackbone=CLIP ViT-B/162026.03 | 82.05 | 73.48 | |
| DualPromptBackbone=CLIP ViT-B/162026.03 | 82.01 | 75.77 | |
| L2PBackbone=CLIP ViT-B/162026.03 | 81.67 | 75.98 | |
| APERBackbone=CLIP ViT-B/162026.03 | 78.62 | 71.35 | |
| CODA-PromptBackbone=CLIP ViT-B/162026.03 | 78 | 67.52 |