RAP: Efficient Text-Video Retrieval with Sparse-and-Correlated Adapter | Read Paper on Bytez