Déjà Vu Memorization in Vision–Language Models | Read Paper on Bytez