- 10 Mar, 2020 1 commit
-
-
Ran Chen authored
In a multi worker set up saving is done on each worker. If they're saving to the same location, e.g. GCS, there will be conflicts. With this change we save to temporary directory on non-chief workers. Note that, there may be synchronization in saving that needs all workers to participate, so we cannot only save on one worker. PiperOrigin-RevId: 300141152
-
- 07 Mar, 2020 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 299594839
-
- 06 Mar, 2020 1 commit
-
-
Zongwei Zhou authored
In BERT SQuAD, disable explicit allreduce for now to keep the original clip_by_global_norm math. With explicit allreduce, the gradients before allreduce are scaled so even if we move clip_by_global_norm before allreduce (as in TF1 and pre-TF 2.2) it will operate on scaled gradients, the math will be changed. So with explicit allreduce, it is better to move clip_by_global_norm to after allreduce. PiperOrigin-RevId: 299278082
-
- 05 Mar, 2020 1 commit
-
-
Zongwei Zhou authored
PiperOrigin-RevId: 299007295
-
- 04 Mar, 2020 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 298717616
-
- 03 Mar, 2020 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 298652484
-
- 02 Mar, 2020 2 commits
-
-
Will Cromar authored
PiperOrigin-RevId: 298466825
-
Hongkun Yu authored
PiperOrigin-RevId: 298402269
-
- 28 Feb, 2020 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 297757527
-
- 26 Feb, 2020 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 297383836
-
- 25 Feb, 2020 2 commits
-
-
Zongwei Zhou authored
PiperOrigin-RevId: 297222995
-
Hongkun Yu authored
PiperOrigin-RevId: 297002741
-
- 20 Feb, 2020 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 296140699
-
- 19 Feb, 2020 2 commits
-
-
Zongwei Zhou authored
PiperOrigin-RevId: 295869937
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 295849975
-
- 13 Feb, 2020 1 commit
-
-
Haitang Hu authored
PiperOrigin-RevId: 294922828
-
- 12 Feb, 2020 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 294560916
-
- 07 Feb, 2020 1 commit
-
-
Haitang Hu authored
This would match the behavior described in on_batch_end() functions in Keras callback. See https://www.tensorflow.org/api_docs/python/tf/keras/callbacks/Callback. PiperOrigin-RevId: 293887583
-
- 27 Jan, 2020 1 commit
-
-
Yanhui Liang authored
PiperOrigin-RevId: 291810091
-
- 16 Jan, 2020 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 290111892
-
- 06 Jan, 2020 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 288388231
-
- 31 Dec, 2019 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 287619398
-
- 04 Dec, 2019 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 283770513
-
- 23 Nov, 2019 1 commit
-
-
Chen Chen authored
PiperOrigin-RevId: 282096004
-
- 18 Nov, 2019 2 commits
-
-
Hongkun Yu authored
PiperOrigin-RevId: 281117886
-
Allen Wang authored
PiperOrigin-RevId: 281063737
-
- 08 Nov, 2019 1 commit
-
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 279361336
-
- 07 Nov, 2019 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 279116219
-
- 06 Nov, 2019 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 278756245
-
- 31 Oct, 2019 1 commit
-
-
Hongkun Yu authored
PiperOrigin-RevId: 277649735
-
- 29 Oct, 2019 1 commit
-
-
A. Unique TensorFlower authored
PiperOrigin-RevId: 277179351
-
- 28 Oct, 2019 2 commits
-
-
Yeqing Li authored
PiperOrigin-RevId: 277156216
-
Hongkun Yu authored
PiperOrigin-RevId: 277087491
-
- 21 Oct, 2019 2 commits
-
-
Hongkun Yu authored
PiperOrigin-RevId: 275879424
-
Yeqing Li authored
PiperOrigin-RevId: 275867562
-
- 18 Oct, 2019 1 commit
-
-
Hongkun Yu authored
Remove train_data_size flag PiperOrigin-RevId: 275545035
-
- 16 Oct, 2019 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 275080469
-
- 15 Oct, 2019 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 274807747
-
- 11 Oct, 2019 1 commit
-
-
A. Unique TensorFlower authored
Change summary directory and model checkpoint directory so that training via Keras Compile/Fit() and custom training loop is consistent. PiperOrigin-RevId: 274202793
-
- 10 Oct, 2019 1 commit
-
-
Yeqing Li authored
PiperOrigin-RevId: 274035928
-