Practical Low-Rank Gradient Compression in Decentralized Deep Learning

NeurIPS 2020