Exploring the limits of Concurrency in ML Training on Google TPUs.
Sameer KumarJames BradburyCliff YoungYu Emma WangAnselm LevskayaBlake A. HechtmanDehao ChenHyoukJoong LeeMehmet DeveciNaveen KumarPankaj KanwarShibo WangSkye Wanderman-MilneSteve LacyTao WangTayo OguntebiYazhou ZuYuanzhong XuAndy SwingPublished in: CoRR (2020)