Loading paper
Self-Calibrated Consistency can Fight Back for Adversarial Robustness in Vision-Language Models | Tomesphere