Image Classification on ObjectNet 1.0 (test)
72.3Top-1 AccuracyCLIP
Evaluation Results
| Method | Links | |
|---|---|---|
| CLIPBackbone=ViT-L/142021.06 | 72.3 | |
| ViT-G/14Backbone=ViT-G/142021.06 | 70.53 | |
| NSBackbone=EfficientNet-L22021.06 | 68.5 | |
| BiT-LBackbone=ResNet2021.06 | 58.7 | |
| RELICv2Epochs=3000, Evaluation=Linear evaluation, Pre-training dataset=JFT-300M2022.01 | 21.3 | |
| RELICv2Epochs=5000, Evaluation=Linear evaluation, Pre-training dataset=JFT-300M2022.01 | 20.9 | |
| RELICv2Epochs=1000, Evaluation=Linear evaluation, Pre-training dataset=JFT-300M2022.01 | 20.3 |