Class-Incremental Learning on TinyImageNet 20 steps 100 base classes (test)
79.32Average AccuracyGIFT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| GIFTBackbone=CLIP2025.03 | 79.32 | 74.87 | |
| ZSCLBackbone=CLIP2025.03 | 77.18 | 68.3 | |
| LwF-VRBackbone=CLIP2025.03 | 69.94 | 63.89 | |
| iCaRLBackbone=CLIP2025.03 | 69.65 | 64.68 | |
| CLIP Zero-shotBackbone=CLIP, Mode=Zero-shot2025.03 | 69.49 | 65.3 | |
| LwFBackbone=CLIP2025.03 | 54.79 | 42.26 | |
| CLIP Fine-tuneBackbone=CLIP, Mode=Fine-tune2025.03 | 54.62 | 44.55 | |
| DyTox2025.03 | 46.18 | 36.21 | |
| UCIR2025.03 | 42.84 | 30.85 | |
| PASS2025.03 | 42.01 | 32.93 | |
| EEIL2025.03 | 40.41 | 29.72 | |
| MUC2025.03 | 21.89 | 10.32 | |
| EWC2025.03 | 12.35 | 4.73 |