General Continual Learning on ImageNet-R
64.23AAUCMISA w/ MePo
Evaluation results
| Method | Links | ||
|---|---|---|---|
| MISA w/ MePoPTM=Sup-21/1K2026.02 | 64.23 | 58.2 | |
| Deep L2P w/ MePoPTM=Sup-21/1K2026.02 | 62.71 | 58.91 | |
| MVP w/ MePoPTM=Sup-21/1K2026.02 | 61.28 | 50.82 | |
| Deep L2P w/ MePoPTM=Sup-21K2026.02 | 58.71 | 55.13 | |
| MISA w/ MePoPTM=iBOT-21K2026.02 | 57 | 49.86 | |
| Deep L2P w/ MePoPTM=iBOT-21K2026.02 | 55.3 | 52.38 | |
| MISA w/ MePoPTM=Sup-21K2026.02 | 54.86 | 49.18 | |
| DualPrompt w/ MePoPTM=Sup-21/1K2026.02 | 54.75 | 44.75 | |
| MVP w/ MePoPTM=iBOT-21K2026.02 | 53.75 | 42.55 | |
| MISAPTM=Sup-21K2026.02 | 51.52 | 45.08 | |
| MVPPTM=Sup-21/1K2026.02 | 51.26 | 41.41 | |
| MISAPTM=Sup-21/1K2026.02 | 50.87 | 47.75 | |
| CODA-PPTM=Sup-21K2026.02 | 50.11 | 46.17 | |
| DualPromptPTM=Sup-21/1K2026.02 | 49.52 | 47.14 | |
| DualPrompt w/ MePoPTM=iBOT-21K2026.02 | 48.06 | 37.69 | |
| Seq FT (SL)PTM=Sup-21K2026.02 | 47.2 | 39.6 | |
| MVP w/ MePoPTM=Sup-21K2026.02 | 46.35 | 38.21 | |
| DualPrompt w/ MePoPTM=Sup-21K2026.02 | 44.65 | 36.76 | |
| MVPPTM=iBOT-21K2026.02 | 43.89 | 34.19 | |
| Deep L2PPTM=Sup-21/1K2026.02 | 42.74 | 39.22 | |
| Deep L2PPTM=Sup-21K2026.02 | 42.39 | 38.16 | |
| MVPPTM=Sup-21K2026.02 | 41.5 | 34.14 | |
| DualPromptPTM=iBOT-21K2026.02 | 41.33 | 35.58 | |
| MISAPTM=iBOT-21K2026.02 | 40.94 | 36.16 | |
| DualPromptPTM=Sup-21K2026.02 | 38.63 | 30.71 | |
| Deep L2PPTM=iBOT-21K2026.02 | 33.68 | 36.24 | |
| Linear ProbePTM=Sup-21K2026.02 | 29.24 | 16.87 | |
| Seq FTPTM=Sup-21K2026.02 | 7.51 | 2.29 |