DeAR: Accelerating Distributed Deep Learning with Fine-Grained All-Reduce Pipelining

Re-assign community