Distilling Out-of-Distribution Robustness from Vision-Language Foundation Models

Open paper
Benchmark Model Rank Results
domain-generalization-on-imagenet-aDiscrete Adversarial Distillation (ViT-B/224)#25Top-1 accuracy %: 31.8
domain-generalization-on-imagenet-aDiscrete Adversarial Distillation (ResNet-50)#32Top-1 accuracy %: 7.7
domain-generalization-on-imagenet-rDiscrete Adversarial Distillation (ViT-B,224)#13Top-1 Error Rate: 34.9
domain-generalization-on-imagenet-sketchDiscrete Adversarial Distillation (ViT-B, 224)#12Top-1 accuracy: 46.1
image-classification-on-imagenetDiscrete Adversarial Distillation (ViT-B, 224)#578Top 1 Accuracy: 81.9%
image-classification-on-imagenet-v2Discrete Adversarial Distillation (ViT-B, 224)#22Top 1 Accuracy: 71.7

Rank counts only results with a code link.