One Prompt Word is Enough to Boost Adversarial Robustness for Pre-trained Vision-Language Models
Primary LanguagePythonMIT LicenseMIT