Distributed deep learning has emerged as an essential approach for training large-scale deep neural networks by utilising multiple computational nodes. This methodology partitions the workload either ...