Combining Global Sparse Gradients with Local Gradients in Distributed Neural Network Training.

Alham Fikri Aji, Kenneth Heafield, Nikolay Bogoychev

Browse the full EMNLP paper archive.