Continual Learning on ImageNet-A 50 tasks
57.08Accuracy LastSR2-LoRA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SR2-LoRABackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 57.08 | 66.47 | |
| LoRA-DRSBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 53.28 | 63.44 | |
| MOSBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 52.29 | 62.91 | |
| SSIATBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 51.44 | 61.63 | |
| SLCABackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 49.31 | 56.72 | |
| MACILBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 47.86 | 59.96 | |
| EASEBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 37.06 | 49.8 | |
| CODA-PromptBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 30.35 | 41.53 |