- 03 11月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 10 9月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 02 9月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 30 8月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 28 8月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Add RAdam, NovoGrad, Lookahead, and AdamW optimizers, a few ResNet tweaks and scheduler factory tweak. * Add some of the trendy new optimizers. Decent results but not clearly better than the standards. * Can create a None scheduler for constant LR * ResNet defaults to zero_init of last BN in residual * add resnet50d config
-
- 13 8月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Add support to split random erasing blocks into randomly selected number with --recount arg. Fix random selection of aspect ratios.
-
- 27 7月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 29 6月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
* reorganize train args * allow resolve_data_config to be used with dict args, not just arparse * stop incrementing epoch before save, more consistent naming vs csv, etc * update resume and start epoch handling to match above * stop auto-incrementing epoch in scheduler
-
- 21 6月, 2019 2 次提交
-
-
由 Ross Wightman 提交于
* Support PyTorch native DDP as fallback if APEX not present * Support SyncBN for both APEX and Torch native (if torch >= 1.1) * EMA model does not appear to need DDP wrapper, no gradients, updated from wrapped model
-
由 Ross Wightman 提交于
-
- 20 6月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 11 6月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Default to img_size in model default_cfg, defer output folder creation until later in the init sequence
-
- 08 6月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
* ModelEma class added to track an EMA set of weights for the model being trained * EMA handling added to train, validation and clean_checkpoint scripts * Add multi checkpoint or multi-model validation support to validate.py * Add syncbn option (APEX) to train script for experimentation * Cleanup interface of CheckpointSaver while adding ema functionality
-
- 29 5月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 26 5月, 2019 1 次提交
-
-
由 Zhun Zhong 提交于
Set input and target to Cuda when without using prefetcher.
-
- 19 5月, 2019 2 次提交
-
-
由 Ross Wightman 提交于
* Do mixup in custom collate fn if prefetcher enabled, reduces performance impact * Move mixup code to own file * Add arg to disable prefetcher * Fix no cuda transfer when prefetcher off * Random erasing when prefetcher off wasn't changed to match new args, fixed * Default random erasing to off (prob = 0.) for train
-
由 Ross Wightman 提交于
-
- 17 5月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
* factor out data related constants to own file * move data related config helpers to own file * add a variant of RandomResizeCrop that randomizes interpolation method * remove old Numpy version of RandomErasing * cleanup torch version of RandomErasing and use it in either GPU loader batch mode or single image cpu Transform
-
- 14 5月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
* initial impl w/ label smoothing converging, but needs more testing
-
- 23 4月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 22 4月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Morph mnasnet impl into a generic mobilenet that covers Mnasnet, MobileNetV1/V2, ChamNet, FBNet, and related * add an alternate RMSprop opt that applies eps like TF * add bn params for passing through alternates and changing defaults to TF style
-
- 18 4月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 13 4月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
* create one resolve fn to pull together model defaults + cmd line args * update attribution comments in some models * test update train/validation/inference scripts
-
- 12 4月, 2019 2 次提交
-
-
由 Ross Wightman 提交于
-
由 Ross Wightman 提交于
* 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
-
- 08 4月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Cleanup CheckpointSaver, add support for increasing or decreasing metric, switch to prec1 metric in train loop
-
- 06 4月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 12 3月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Improve creation of data pipeline with prefetch enabled vs disabled, fixup inception_res_v2 and dpn models
-
- 11 3月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
Add some Nvidia performance enhancements (prefetch loader, fast collate), and refactor some of training and model fact/transforms
-
- 02 3月, 2019 2 次提交
-
-
由 Ross Wightman 提交于
-
由 Ross Wightman 提交于
-
- 23 2月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 18 2月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 14 2月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 09 2月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 07 2月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 03 2月, 2019 1 次提交
-
-
由 Ross Wightman 提交于
-
- 02 2月, 2019 2 次提交
-
-
由 Ross Wightman 提交于
-
由 Ross Wightman 提交于
-