Image Classification on ImageNet Real labels (test)
90.9Top-1 AccuracyFixEfficientNet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| FixEfficientNetExtra Training Data=true, Model Scale=L22020.03 | 90.9 | 98.8 | — | |
| Vision Transformer (ViT-H/14)Pre-training Dataset=JFT-300M, Architecture=ViT-H/142020.10 | 90.72 | — | — | |
| EfficientNetExtra Training Data=true, Model Scale=L22020.03 | 90.6 | 98.8 | — | |
| Noisy StudentArchitecture=EfficientNet-L22020.10 | 90.55 | — | — | |
| Vision Transformer (ViT-L/16)Pre-training Dataset=JFT-300M, Architecture=ViT-L/162020.10 | 90.54 | — | — | |
| BiT-LArchitecture=ResNet152x42020.10 | 90.54 | — | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B72020.03 | 90.3 | 98.7 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B62020.03 | 90.1 | 98.6 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B72020.03 | 90.1 | 98.6 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B52020.03 | 90 | 98.6 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B82020.03 | 90 | 98.6 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B42020.03 | 89.8 | 98.5 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B62020.03 | 89.8 | 98.5 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B52020.03 | 89.7 | 98.5 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B72020.03 | 89.7 | 98.5 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B62020.03 | 89.6 | 98.4 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B82020.03 | 89.6 | 98.3 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B42020.03 | 89.4 | 98.4 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B52020.03 | 89.4 | 98.4 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B72020.03 | 89.4 | 98.3 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B62020.03 | 89.3 | 98.3 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B32020.03 | 89.2 | 98.4 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B42020.03 | 89.2 | 98.3 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B52020.03 | 88.9 | 98.2 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B22020.03 | 88.8 | 98.2 | — | |
| Vision Transformer (ViT-L/16)Pre-training Dataset=ImageNet-21k, Architecture=ViT-L/162020.10 | 88.62 | — | — | |
| EfficientNetExtra Training Data=true, Model Scale=B32020.03 | 88.4 | 98 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B32020.03 | 88.3 | 98 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B42020.03 | 88.3 | 97.9 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B12020.03 | 88.1 | 98 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B22020.03 | 87.7 | 97.6 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B22020.03 | 87.3 | 97.6 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B32020.03 | 87.2 | 97.4 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B12020.03 | 87 | 97.4 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B12020.03 | 86.7 | 97.2 | — | |
| FixEfficientNetExtra Training Data=true, Model Scale=B02020.03 | 86.5 | 97.3 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B22020.03 | 86 | 96.8 | — | |
| FixEfficientNetExtra Training Data=false, Model Scale=B02020.03 | 85.8 | 96.8 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B12020.03 | 85.1 | 96.4 | — | |
| EfficientNetExtra Training Data=true, Model Scale=B02020.03 | 84.5 | 96.4 | — | |
| EfficientNetExtra Training Data=false, Model Scale=B02020.03 | 83.7 | 95.8 | — | |
| Meta Pseudo LabelsBackbone=EfficientNet-L22020.03 | — | — | 91.02 | |
| Previous SOTABackbone=EfficientNet-L22020.03 | — | — | 90.72 |