Critical Parameters for Scalable Distributed Learning with Large Batches and Asynchronous Updates

Open in new window