Continual Learning on Split CUB-200 20 Tasks
94.32Average AccuracySR2-LoRA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SR2-LoRABackbone=ViT-B/16-IN21K2026.05 | 94.32 | — | — | 89.71 | |
| MOSBackbone=ViT-B/16-IN21K2026.05 | 93.66 | — | — | 89.42 | |
| SSIATBackbone=ViT-B/16-IN21K2026.05 | 93.52 | — | — | 89.06 | |
| MACILBackbone=ViT-B/16-IN21K2026.05 | 93.52 | — | — | 88.63 | |
| LoRA-DRSBackbone=ViT-B/16-IN21K2026.05 | 92.66 | — | — | 87.5 | |
| EASEBackbone=ViT-B/16-IN21K2026.05 | 91.02 | — | — | 84.51 | |
| SLCABackbone=ViT-B/16-IN21K2026.05 | 90.14 | — | — | 82.48 | |
| CODA-PromptBackbone=ViT-B/16-IN21K2026.05 | 78.1 | — | — | 66.41 | |
| Adaptive Trainingsequence_length=20 tasks2026.02 | 36.24 | 72.59 | 39.6 | — | |
| Adaptive Training2026.02 | 36.24 | 72.59 | 39.6 | — | |
| Adaptive Training2026.02 | 36.24 | 72.59 | 39.6 | — | |
| Frozen Trainingsequence_length=20 tasks2026.02 | 34.9 | 69.21 | 36.53 | — | |
| Frozen Training2026.02 | 34.9 | 69.21 | 36.53 | — | |
| Frozen Training2026.02 | 34.9 | 69.21 | 36.53 | — | |
| Initialized Trainingsequence_length=20 tasks2026.02 | 34.37 | 71.55 | 40.45 | — | |
| Initialized Training2026.02 | 34.37 | 71.55 | 40.45 | — | |
| Initialized Training2026.02 | 34.37 | 71.55 | 40.45 | — |