List your company
ResearchFormerly WizWand

Adversarial Zero-shot Image Classification on ImageNet (val)

61.7Adversarial AccuracyTAME
-2.46814.19130.8547.509May 17, 2026
Updated 5mo ago

Evaluation results

MethodLinks
TAMEAttack Method=PGD, Backbone=ViT-B/16, Input Configuration=V-L Independent2026.0561.7
TAMEAttack Method=AutoAttack, Backbone=ViT-B/16, Input Configuration=V-L Independent2026.0560.8
TAMEAttack Method=CW, Backbone=ViT-B/16, Input Configuration=V-L Independent2026.0560.7
TAMEAttack Method=DI, Backbone=ViT-B/16, Input Configuration=V-L Independent2026.0559.4
R-TPTAttack Method=PGD, Backbone=ViT-B/162026.0554.8
TAPTAttack Method=PGD, Backbone=ViT-B/16, Input Configuration=V-L Independent2026.0550
MMoPAttack Method=PGD, Backbone=ViT-B/16, Input Configuration=V-L Independent2026.0528
NAP-TuningAttack Method=PGD, Backbone=ViT-B/162026.0527.2
APT-VLIAttack Method=PGD, Backbone=ViT-B/162026.0524.3
APT-VLJAttack Method=PGD, Backbone=ViT-B/162026.0523.9
FAPAttack Method=PGD, Backbone=ViT-B/162026.0523.2
APT-VAttack Method=PGD, Backbone=ViT-B/162026.0519.4
VanillaAttack Method=DI, Backbone=ViT-B/162026.056.1
APT-TAttack Method=PGD, Backbone=ViT-B/162026.051.5
VanillaAttack Method=PGD, Backbone=ViT-B/162026.051.4
VanillaAttack Method=CW, Backbone=ViT-B/162026.050.2
VanillaAttack Method=AutoAttack, Backbone=ViT-B/162026.050