Targeted Forgetting of Image Subgroups in CLIP Models | Read Paper on Bytez