IBM speeds deep learning by using multiple servers

IBM's Distributed Deep Learning spreads model training across any number of hardware nodes—as long as they’re IBM nodes