降低分布式训练通信的梯度稀疏压缩方法
陈世达,刘强,韩亮

Gradient sparsification compression approach to reducing communication in distributed training
Shi-da CHEN,Qiang LIU,Liang HAN
图 1 ResNet50累计梯度统计直方图
Fig.1 Histogram of residual gradient statistics in ResNet50