Classification on Tiny-ImageNet
88.6Baseline Mean TAROKA
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| ROKAModel=ViT-base/16, Unlearning Method=Targeted, Task=Classification2026.02 | 88.6 | 87.31 | 19.4 | 86.57 | |
| ROKAModel=ViT-base/16, Unlearning Method=Non-Targeted, Task=Classification2026.02 | 82.99 | 90.71 | 21.2 | 89.66 | |
| ROKAModel=DeiT-tiny/16, Unlearning Method=Non-Targeted, Task=Classification2026.02 | 76.6 | 67.84 | 50.8 | 63.21 | |
| ROKAModel=DeiT-tiny/16, Unlearning Method=Targeted, Task=Classification2026.02 | 73.2 | 67.58 | 1.2 | 64.97 | |
| ROKAModel=CLIP-large/14, Unlearning Method=Targeted, Task=Zero-shot Classification2026.02 | 68.39 | 67.04 | 1 | 81.02 | |
| ROKAModel=CLIP-large/14, Unlearning Method=Non-Targeted, Task=Zero-shot Classification2026.02 | 65.81 | — | 12 | 63.81 | |
| ROKAModel=CLIP-base/32, Unlearning Method=Targeted, Task=Zero-shot Classification2026.02 | 46.79 | 49.49 | 0 | 44.73 | |
| ROKAModel=CLIP-base/32, Unlearning Method=Non-Targeted, Task=Zero-shot Classification2026.02 | 42.2 | — | 8 | 43.91 |