Improving CLIP Robustness with Knowledge Distillation and Self-Training

Open in new window