CLAP4CLIP: Continual Learning with Probabilistic Finetuning for Vision-Language Models | Read Paper on Bytez