GradSA: Gradient Sparsification and Accumulation for Communication-Efficient Distributed Deep Learning.

Bo Liu, Wenbin Jiang, Shaofeng Zhao, Hai Jin, Bingsheng He

Browse the full GPC paper archive.