The Case for Strong Scaling in Deep Learning: Training Large 3D CNNs With Hybrid Parallelism.
Yosuke OyamaNaoya MaruyamaNikoli DrydenErin McCarthyPeter HarringtonJan BalewskiSatoshi MatsuokaPeter NugentBrian Van EssenPublished in: IEEE Trans. Parallel Distributed Syst. (2021)