Unlocking the Capabilities of Large Vision-Language Models for Generalizable and Explainable Deepfake Detection | Read Paper on Bytez