Image Classification on ImageNet V2 1 (val)
73.9Top-1 AccuracyHAT (+KD)
Evaluation Results
| Method | Links | |
|---|---|---|
| HAT (+KD)Backbone=ViT-B, Training Set=ImageNet-1K, Knowledge Distillation=true2022.04 | 73.9 | |
| DeiT (+KD)Backbone=ViT-B, Training Set=ImageNet-1K, Knowledge Distillation=true2022.04 | 73.2 | |
| HATBackbone=ViT-B, Training Set=ImageNet-1K2022.04 | 72.6 | |
| DeiTBackbone=ViT-B, Training Set=ImageNet-1K2022.04 | 71.5 | |
| PyramidATBackbone=ViT-B, Training Set=ImageNet-1K2022.04 | 70.8 | |
| SAMBackbone=ViT-B, Training Set=ImageNet-1K2022.04 | 67.5 | |
| VanillaBackbone=ViT-B, Training Set=ImageNet-1K2022.04 | 64.1 |