Session <Full Schedule · Contributors · Organizations · Search Program · My Schedule · Happening Now · MapsMore…Search ProgramMy ScheduleHappening NowMapsPaper: Training in HPC Machine LearningSession ChairShuaiwen Leon SongMicrosoft CorporationEvent TypePaperTimeWednesday, 15 November 20231:30pm - 3pm MSTLocation401-402Tags Artificial Intelligence/Machine Learning Registration Categories TP Presentations1:30pm - 2:00pm MSTEasyScale: Elastic Training with Consistent Accuracy and Improved Utilization on GPUsAuthorsMingzhen LiWencong XiaoHailong YangBiao SunHanyu ZhaoShiru RenZhongzhi LuanXianyan JiaYi LiuYong LiWei LinDepei Qian2:00pm - 2:30pm MSTHanayo: Harnessing Wave-Like Pipeline Parallelism for Enhanced Large Model Training EfficiencyAuthorsZiming LiuShenggan ChengHaotian ZhouYang You2:30pm - 3:00pm MSTHigh Throughput Training of Deep Surrogates from Large Ensemble RunsAuthorsLucas MeyerMarc SchoulerRobert Alexander CaulkAlejandro RibesBruno Raffin