Vincent Wang 28e4c0edf5 Keep loss_scale and Whole Loss Subgraph in FP32 during Mixed Precision Training (#4268) ... * Keep loss subgraph as FP32 when mixed-p training. * Fix case where there is no white-list loss op. * Get nodes from loss_scale instead of whitelist. * rename const variables. Co-authored-by: Vincent Wang <weicwang@OrtDevTest2v100.af05slrtruoetgaxwwjv5nsq5e.px.internal.cloudapp.net>		2020-07-03 06:54:56 +08:00
..
orttraining	Keep loss_scale and Whole Loss Subgraph in FP32 during Mixed Precision Training (#4268)	2020-07-03 06:54:56 +08:00
pytorch_frontend_examples	Remove ORTModel from frontend API (#3825)	2020-05-11 18:20:33 -07:00
tools	adjust bs size (#4375)	2020-06-30 10:29:48 -07:00