Image Classification on ImageNet-9 (IN-9 Gap)
-2.35IN-9 GapMAE+LLE
Evaluation results
| Method | Links | |
|---|---|---|
| MAE+LLEtrain data=IN-1k, evaluation protocol=Last Layer Ensemble, backbone=ViT-H2022.12 | -2.35 | |
| SWAGtrain data=IG-3.6B, evaluation protocol=fine-tuning, backbone=ViT-H2022.12 | -3.19 | |
| MAEtrain data=IN-1k, evaluation protocol=fine-tuning, backbone=ViT-H2022.12 | -3.24 | |
| CLIPtrain data=LAION-2B, evaluation protocol=zero-shot, backbone=ViT-H2022.12 | -3.71 | |
| SWAGtrain data=IG-3.6B, evaluation protocol=linear probing, backbone=ViT-H2022.12 | -7.86 |