FLuID: Mitigating Stragglers in Federated Learning using Invariant Dropout | Read Paper on Bytez