Fetching the paper…

Variance-based Gradient Compression for Efficient Distributed Deep Learning · Around