Quantifying Cross-Modality Memorization in Vision-Language Models | Read Paper on Bytez