Continual Learning on Split CUB (test)
73.8AT (%)MULTI-TASK
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MULTI-TASKModel variant=joint embedding2018.12 | 73.8 | — | — | |
| A-GEMModel variant=joint embedding2018.12 | 71 | 0.07 | 0.54 | |
| EWCModel variant=joint embedding2018.12 | 68.4 | 0.09 | 0.52 | |
| RWALKModel variant=joint embedding2018.12 | 67.4 | 0.1 | 0.52 | |
| VANModel variant=joint embedding2018.12 | 67.1 | 0.1 | 0.52 | |
| PIModel variant=joint embedding2018.12 | 66.6 | 0.1 | 0.52 | |
| MULTI-TASKModel variant=standard2018.12 | 65.6 | — | — | |
| A-GEMModel variant=standard2018.12 | 62 | 0.07 | 0.3 | |
| PIModel variant=standard2018.12 | 55.3 | 0.12 | 0.29 | |
| RWALKModel variant=standard2018.12 | 54.4 | 0.13 | 0.29 | |
| VANModel variant=standard2018.12 | 54.3 | 0.13 | 0.29 | |
| EWCModel variant=standard2018.12 | 54 | 0.13 | 0.29 |