Fetching the paper…

Poseidon: An Efficient Communication Architecture for Distributed Deep Learning on GPU Clusters · Around