Image Classification on Generalization Suite Unseen: RESISC45, SVHN ViT-B/32 (Split 2)
86.7Accuracy (Cars)TARA-Variant B
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TARA-Variant BBackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 86.7 | 77 | 79.5 | 42.8 | 76.3 | 98.2 | 76.8 | 70.5 | 45.6 | 58.1 | 72.1 | |
| TARA-Variant ABackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 84.9 | 75.7 | 76.3 | 40.5 | 77.4 | 97.5 | 75.4 | 69.5 | 42.1 | 55.8 | 70.5 | |
| KnOTS-DARE-TIESBackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 82.7 | 72.2 | 53.1 | 48.5 | 61.5 | 94.3 | 68.7 | 66.7 | 35.4 | 51 | 64.3 | |
| KnOTS-TIESBackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 82.5 | 72.9 | 53.8 | 47.8 | 61.7 | 94.9 | 69 | 67.5 | 35.1 | 51.3 | 64.5 | |
| TABackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 82.3 | 75.1 | 53 | 40.7 | 52.4 | 96.7 | 66.7 | 68.6 | 34.7 | 51.7 | 63 | |
| TIESBackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 81.9 | 74.8 | 73.7 | 33.2 | 63.4 | 95.6 | 70.4 | 68.9 | 31.6 | 50.3 | 65.4 | |
| LoRA-LEGOBackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 81.1 | 72.3 | 55.6 | 42.8 | 62.6 | 94.7 | 68.2 | 65.2 | 33.9 | 49.5 | 63.5 | |
| AdaMergingBackbone=ViT-B/32, Normalization=to finetuned (%)2026.03 | 79.6 | 74.2 | 72.5 | 36.5 | 60 | 97.9 | 70.1 | 69.2 | 41.1 | 55.2 | 66.4 |