Continual Learning on ImageNet-R 50 tasks
74.85Accuracy (Last Task)SR2-LoRA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| SR2-LoRABackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 74.85 | — | — | 80.51 | |
| SSIATBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 74.52 | — | — | 79.01 | |
| LoRA-DRSBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 72.17 | — | — | 78.05 | |
| MACILBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 70.1 | — | — | 77.47 | |
| SLCABackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 68.95 | — | — | 71.15 | |
| EASEBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 68.53 | — | — | 75.49 | |
| CL-LoRABackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 68.36 | — | — | 76.8 | |
| MOSBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 66.92 | — | — | 74.81 | |
| InfLoRABackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 61.91 | — | — | 71.2 | |
| CODA-PromptBackbone=ViT-B/16-IN21K, Number of Tasks=502026.05 | 48.89 | — | — | 55.59 | |
| ARCL-ViTVenue=This work2026.02 | — | 76.45 | 8.2 | — | |
| ConvPromptVenue=CVPR’242026.02 | — | 64.61 | 7.12 | — | |
| CPromptVenue=CVPR’242026.02 | — | 70.75 | 7.44 | — | |
| EASEVenue=CVPR’242026.02 | — | 70.27 | 6.73 | — | |
| InfLoRAVenue=CVPR’242026.02 | — | 62.81 | 10.37 | — | |
| L2PVenue=CVPR’222026.02 | — | 51.38 | 12.34 | — | |
| OVOR-DeepVenue=ICLR’242026.02 | — | 63.25 | 5.23 | — | |
| Seq-FTVenue=Baseline2026.02 | — | 22.57 | 70.01 | — | |
| VPT-CPGVenue=AAAI’252026.02 | — | 73.08 | 12.12 | — |