Commit Graph

8 Commits (9c3859fb9c5bebd224c2f275b16fe0307300623c)

Author SHA1 Message Date
Ross Wightman 9c3859fb9c Uniform pretrained model handling.
* All models have 'default_cfgs' dict
* load/resume/pretrained helpers factored out
* pretrained load operates on state_dict based on default_cfg
* test all models in validate
* schedule, optim factor factored out
* test time pool wrapper applied based on default_cfg
2019-04-11 21:32:16 -07:00
Ross Wightman 63e677d03b Merge branch 'master' of github.com:rwightman/pytorch-models 2019-04-10 14:55:54 -07:00
Ross Wightman 0bc50e84f8 Lots of refactoring and cleanup.
* Move 'test time pool' to Module that can be used by any model, remove from DPN
* Remove ResNext model file and combine with ResNet
* Remove fbresnet200 as it was an old conversion and pretrained performance not worth param count
* Cleanup adaptive avgmax pooling and add back conctat variant
* Factor out checkpoint load fn
2019-04-10 14:53:34 -07:00
Ross Wightman c0e6e5f3db Add common model interface to pnasnet and xception, update factory 2019-04-06 13:59:15 -07:00
Ross Wightman 5180f94c7e Distributed (multi-process) train, multi-gpu single process train, and NVIDIA AMP support 2019-04-05 10:53:04 -07:00
Ross Wightman 5cb1a35c6b Fixup Resnext, remove alternate shortcut types 2019-04-01 11:03:37 -07:00
Ross Wightman 45cde6f0c7 Improve creation of data pipeline with prefetch enabled vs disabled, fixup inception_res_v2 and dpn models 2019-03-11 22:17:42 -07:00
Ross Wightman 2295cf56c2 Add some Nvidia performance enhancements (prefetch loader, fast collate), and refactor some of training and model fact/transforms 2019-03-10 14:23:16 -07:00