Federated Disentangled Tuning with Textual Prior Decoupling and Visual Dynamic Adaptation | Read Paper on Bytez