Image Classification on ImageNet 1000
64.12Top-1 AccuracySUPERVISED
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SUPERVISEDEvaluation Protocol=Linear probe, Backbone=ResNet-50, Epochs=100, Batch Size=5122026.01 | 64.12 | 84.6 | |
| SUPERVISEDEvaluation Protocol=k-NN, K=200, Backbone=ResNet-50, Epochs=100, Batch Size=5122026.01 | 61.11 | 85.06 | |
| HypersolidEvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 61.11 | 84.31 | |
| DINOEvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 58.49 | 82.42 | |
| BTEvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 57.51 | 81.06 | |
| VICRegEvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 57.47 | 80.81 | |
| SimCLREvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 55.47 | 79.88 | |
| BYOLEvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 54.07 | 78.77 | |
| LeJEPAEvaluation Protocol=Linear probe, Backbone=ResNet-502026.01 | 53.5 | 77.7 | |
| HypersolidEvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 50.65 | 78.15 | |
| DINOEvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 49.72 | 78.06 | |
| VICRegEvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 45.26 | 73.23 | |
| BTEvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 45.03 | 73.2 | |
| SimCLREvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 41.47 | 70.06 | |
| BYOLEvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 36.8 | 65.05 | |
| LeJEPAEvaluation Protocol=k-NN, K=200, Backbone=ResNet-502026.01 | 32.52 | 60.81 |