List your company
ResearchFormerly WizWand

Semi-supervised Classification on ImageNet 10% labels (train)

0.765Top-1 AccuracyTWIST
0.555960.610230.66450.71877Oct 14, 2021
Updated 8mo ago

Evaluation results

MethodLinks
TWISTBackbone=ViT-B/162021.100.7650.931
TWISTBackbone=ResNet-50x22021.100.7530.928
DINOBackbone=ViT-B/162021.100.7460.92
BYOLBackbone=ResNet-50x22021.100.7350.917
TWISTBackbone=ResNet-502021.100.7170.91
SimCLRBackbone=ResNet-50x22021.100.7170.912
SwAVBackbone=ResNet-502021.100.7020.899
BarlowTwinsBackbone=ResNet-502021.100.6970.893
BYOLBackbone=ResNet-502021.100.6880.89
DINOBackbone=ResNet-502021.100.6820.891
SimCLRBackbone=ResNet-502021.100.6560.878
SUPBackbone=ResNet-502021.100.5640.804