MiCRO: Near-Zero Cost Gradient Sparsification for Scaling and Accelerating Distributed DNN Training | IEEE Conference Publication | IEEE Xplore