List your company
ResearchFormerly WizWand

Image Classification on ImageNet1K 1.0 (train val)

80.4Top-1 AccMR BarTwins
55.75262.15168.5574.949Nov 15, 2022
Updated 8mo ago

Evaluation results

MethodLinks
MR BarTwinsArch=ViT-B, Params.(M)=113, Pretraining-epochs=600, Supervision=RGB, α=7, λ=22022.1180.4—————
DINOArch=ViT-B, Params.(M)=88, Pretraining-epochs=300, Supervision=RGB2022.1180.1—————
MR SimCLRArch=ViT-B, Params.(M)=113, Pretraining-epochs=600, Supervision=RGB, α=8, λ=52022.1180—————
iBOTArch=ViT-B, Params.(M)=88, Pretraining-epochs=1600, Supervision=RGB2022.1179.5—————
MoCo-v3Arch=ViT-B, Params.(M)=88, Pretraining-epochs=600, Supervision=RGB2022.1176.5—————
MAEArch=ViT-L, Params.(M)=218, Pretraining-epochs=1600, Supervision=RGB2022.1175.1—————
BYOLArch=ResNet-50, Params.(M)=23, Pretraining-epochs=1000, Supervision=RGB2022.1174.3—————
BarTwins(w/ ViT)Arch=ViT-B, Params.(M)=87, Pretraining-epochs=600, Supervision=RGB2022.1174.1—————
BarTwinsArch=ResNet-50, Params.(M)=23, Pretraining-epochs=1000, Supervision=RGB2022.1173.2—————
SimCLR(w/ ViT)Arch=ViT-B, Params.(M)=87, Pretraining-epochs=600, Supervision=RGB2022.1173.1—————
InfoMinArch=ResNet-50, Params.(M)=23, Pretraining-epochs=800, Supervision=RGB2022.1173—————
MoCo-v2Arch=ResNet-50, Params.(M)=23, Pretraining-epochs=800, Supervision=RGB2022.1171.1—————
SimCLRArch=ResNet-50, Params.(M)=23, Pretraining-epochs=1000, Supervision=RGB2022.1169.3—————
CAEArch=ViT-B, Params.(M)=88, Pretraining-epochs=600, Supervision=DALLE+RGB2022.1168.3—————
MaskFeatArch=ViT-L, Params.(M)=218, Pretraining-epochs=1600, Supervision=HOG2022.1167.7—————
InfoCLArch=ResNet-50, Params.(M)=23, Pretraining-epochs=200, Supervision=RGB2022.1161.6—————
BEITArch=ViT-B, Params.(M)=86, Pretraining-epochs=800, Supervision=RGB2022.1156.7—————
SimMIMArch=ViT-B, Params.(M)=88, Pretraining-epochs=800, Supervision=RGB2022.1156.7—————
BaselineModel=DeiT-tiny, Hyper?=-, Architecture?=-, Strategy?=-2024.03—72.2—3,30054.61
BaselineModel=DeiT-small, Hyper?=-, Architecture?=-, Strategy?=-2024.03—79.8—13,000124.51
BaselineModel=DeiT-base, Hyper?=-, Architecture?=-, Strategy?=-2024.03—81.8—52,000292.81
NetworkExpansion 6→12Model=DeiT-tiny, Hyper?=✓, Architecture?=✓, Strategy?=x2024.03—70.370.12,50043.21.26
NetworkExpansion 6→12Model=DeiT-small, Hyper?=✓, Architecture?=✓, Strategy?=x2024.03—78.878.69,800100.31.24
NetworkExpansion 6→12Model=DeiT-base, Hyper?=✓, Architecture?=✓, Strategy?=x2024.03—8181.539,000226.81.29
S2ViTE-TinyModel=DeiT-tiny, Hyper?=x, Architecture?=x, Strategy?=✓2024.03—70.1—2,500—1.19
StackBERTModel=DeiT-base, Hyper?=✓, Architecture?=✓, Strategy?=x2024.03—80.8—42,000231.61.26
ToE Hyper r1=0.4Model=DeiT-base, Hyper?=x, Architecture?=✓, Strategy?=✓2024.03—81.7—33,000202.81.44
ToE Hyper r1=0.5Model=DeiT-base, Hyper?=x, Architecture?=✓, Strategy?=✓2024.03—81.8—36,000213.21.37
ToE r1=0.4Model=DeiT-base, Hyper?=✓, Architecture?=✓, Strategy?=✓2024.03—81.4—38,000225.21.3
ToE r1=0.5Model=DeiT-tiny, Hyper?=✓, Architecture?=✓, Strategy?=✓2024.03—72.6—2,60044.21.24
ToE r1=0.5Model=DeiT-small, Hyper?=✓, Architecture?=✓, Strategy?=✓2024.03—79.8—10,000102.21.22
ToE r1=0.5Model=DeiT-base, Hyper?=✓, Architecture?=✓, Strategy?=✓2024.03—81.6—40,000231.21.27
ToMeDeiT r8→Model=DeiT-tiny, Hyper?=✓, Architecture?=x, Strategy?=✓2024.03—71.7—2,50053.31.02
ToMeDeiT r8→Model=DeiT-small, Hyper?=✓, Architecture?=x, Strategy?=✓2024.03—79.7—9,800121.51.02