Image Classification on CIFAR-100, SVHN, Food-101, ImageNet-R Average
87.4Mean AccuracyAdaRoute
Evaluation results
| Method | Links | |
|---|---|---|
| AdaRoute# P (M)=3.9, Backbone=ViT-B/162026.02 | 87.4 | |
| Mona# P (M)=3.9, Backbone=ViT-B/162026.02 | 86.7 | |
| GPS# P (M)=3.5, Backbone=ViT-B/162026.02 | 86.6 | |
| SPT-LoRA# P (M)=3.7, Backbone=ViT-B/162026.02 | 86.5 | |
| SNELL# P (M)=3.7, Backbone=ViT-B/162026.02 | 86.3 | |
| RepAdapter# P (M)=4.3, Backbone=ViT-B/162026.02 | 85.6 | |
| LoRA# P (M)=3.9, Backbone=ViT-B/162026.02 | 85.1 | |
| AdaptFormer# P (M)=4.3, Backbone=ViT-B/162026.02 | 85.1 | |
| LoRand# P (M)=3.9, Backbone=ViT-B/162026.02 | 85.1 | |
| Full-tuning# P (M)=86.0, Backbone=ViT-B/162026.02 | 84.4 | |
| VPT# P (M)=0.1, Backbone=ViT-B/162026.02 | 81.3 |