List your company
ResearchFormerly WizWand

Vision Model Representation Alignment on ImageNet 20% held-out split of 20,000 samples (val)

85Reconstruction AccuracyMOT + R
12.09631.02349.9568.877Oct 2, 2025
Updated 6mo ago

Evaluation results

MethodLinks
MOT + RModel 1=ViT-MAE Base, Model 2=ViT-MAE Large2025.1085
MOT + RModel 1=DINOv2 Small, Model 2=DINOv2 Giant2025.1079
MOT + RModel 1=ViT-MAE Base, Model 2=ViT-MAE Huge2025.1078.8
MOT + RModel 1=DINOv2 Small, Model 2=DINOv2 Large2025.1077.8
MOT + RModel 1=ViT-MAE Base, Model 2=DINOv2 Large2025.1073.2
MOT + RModel 1=DINOv2 Small, Model 2=ViT-MAE Huge2025.1065.7
MOT + RModel 1=DINOv2 Small, Model 2=ViT-MAE Large2025.1063.3
Pairwise Best OTModel 1=ViT-MAE Base, Model 2=DINOv2 Large2025.1062.4
MOT + RModel 1=ViT-MAE Huge, Model 2=DINOv2 Giant2025.1061.4
MOT + RModel 1=DINOv2 Small, Model 2=ViT-MAE Base2025.1060
Pairwise Best OTModel 1=ViT-MAE Base, Model 2=ViT-MAE Large2025.1059.8
Pairwise Best + RModel 1=ViT-MAE Base, Model 2=ViT-MAE Large2025.1059.6
MOTModel 1=ViT-MAE Base, Model 2=ViT-MAE Large2025.1058.8
MOT + RModel 1=ViT-MAE Base, Model 2=DINOv2 Giant2025.1058
MOTModel 1=ViT-MAE Base, Model 2=DINOv2 Large2025.1057.7
Pairwise Best + RModel 1=ViT-MAE Base, Model 2=ViT-MAE Huge2025.1057.1
Pairwise Best + RModel 1=DINOv2 Small, Model 2=ViT-MAE Base2025.1052.6
Pairwise Best + RModel 1=DINOv2 Small, Model 2=ViT-MAE Large2025.1050.9
Pairwise Best + RModel 1=DINOv2 Small, Model 2=ViT-MAE Huge2025.1050.8
MOTModel 1=DINOv2 Small, Model 2=DINOv2 Giant2025.1046.6
Pairwise Best OTModel 1=DINOv2 Small, Model 2=DINOv2 Giant2025.1043.3
Pairwise Best + RModel 1=DINOv2 Small, Model 2=DINOv2 Giant2025.1041.8
Pairwise Best OTModel 1=ViT-MAE Base, Model 2=ViT-MAE Huge2025.1041.7
MOTModel 1=DINOv2 Small, Model 2=ViT-MAE Huge2025.1041.1
Pairwise Best + RModel 1=DINOv2 Small, Model 2=DINOv2 Large2025.1039.4
Pairwise Best OTModel 1=DINOv2 Small, Model 2=ViT-MAE Huge2025.1038.6
MOTModel 1=DINOv2 Small, Model 2=ViT-MAE Large2025.1038.1
Pairwise Best + RModel 1=ViT-MAE Huge, Model 2=DINOv2 Giant2025.1035.9
Pairwise Best OTModel 1=DINOv2 Small, Model 2=ViT-MAE Large2025.1035.4
MOTModel 1=DINOv2 Small, Model 2=DINOv2 Large2025.1035.3
Pairwise Best OTModel 1=ViT-MAE Huge, Model 2=DINOv2 Giant2025.1035.2
Pairwise Best OTModel 1=DINOv2 Small, Model 2=DINOv2 Large2025.1034
MOTModel 1=ViT-MAE Huge, Model 2=DINOv2 Giant2025.1031.7
Pairwise Best OTModel 1=DINOv2 Small, Model 2=ViT-MAE Base2025.1030.1
Pairwise Best + RModel 1=ViT-MAE Base, Model 2=DINOv2 Giant2025.1029.3
MOTModel 1=DINOv2 Small, Model 2=ViT-MAE Base2025.1028.9
Pairwise Best + RModel 1=ViT-MAE Base, Model 2=DINOv2 Large2025.1028.3
MOTModel 1=ViT-MAE Base, Model 2=DINOv2 Giant2025.1020.2
Pairwise Best OTModel 1=ViT-MAE Base, Model 2=DINOv2 Giant2025.1018
MOTModel 1=ViT-MAE Base, Model 2=ViT-MAE Huge2025.1014.9