3 Commits

Author	SHA1	Message	Date
Simon Layton	2bfacff426	Fp16 training initializers ... Summary: Adds support for generating and training pfp16 models. Added SGD optimizer for multi-precision trainers and a new callback to data_parallel_model in order to help multi-precision models keep their different copies of parameters in sync during training. Closes https://github.com/caffe2/caffe2/pull/697 Differential Revision: D5159712 Pulled By: salexspb fbshipit-source-id: 60a889494d2e2f4df1d720331e19f638c5eb95cc	2017-05-31 17:46:58 -07:00
Simon Layton	2c3071fc4e	Rework initializers to pass a class not object ... Summary: Changed tests Moved to WeightInitializer, BiasInitializer keywords Closes https://github.com/caffe2/caffe2/pull/682 Reviewed By: Yangqing Differential Revision: D5138769 Pulled By: salexspb fbshipit-source-id: 81d266100b2a95c64c0196c16670dfd34ea03e02	2017-05-30 09:06:56 -07:00
Alexander Sidorov	016f72537a	ModelHelper.create_param, Initializer abstraction and ParameterInfo for optimizers ... Summary: This is going to unblock Nvidia in their work on adding fp16 support to Caffe2. I discussed this with kennyhorror before to make sure this fits into his work on parameter sharing. Reviewed By: kennyhorror Differential Revision: D5127797 fbshipit-source-id: 4db155d320b1862570c23b77c4252bdacbf2296f	2017-05-25 22:03:15 -07:00