- 12 3月, 2020 5 次提交
-
-
由 Will Cromar 提交于
PiperOrigin-RevId: 300433601
-
由 Yeqing Li 提交于
PiperOrigin-RevId: 300426073
-
由 Jose Baiocchi 提交于
PiperOrigin-RevId: 300399639
-
由 Allen Wang 提交于
PiperOrigin-RevId: 300362532
-
由 Ruoxin Sang 提交于
PiperOrigin-RevId: 300351756
-
- 11 3月, 2020 6 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 300247220
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 300222585
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 300203487
-
由 Ran Chen 提交于
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
-
由 Ruoxin Sang 提交于
PiperOrigin-RevId: 300127838
-
由 Chen Chen 提交于
PiperOrigin-RevId: 300115297
-
- 10 3月, 2020 3 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 300009545
-
由 Rick Chao 提交于
PiperOrigin-RevId: 299998039
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 299901483
-
- 09 3月, 2020 4 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 299792626
-
由 Zongwei Zhou 提交于
PiperOrigin-RevId: 299764679
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 299759804
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 299757650
-
- 08 3月, 2020 1 次提交
-
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 299594839
-
- 07 3月, 2020 1 次提交
-
-
由 Rick Chao 提交于
PiperOrigin-RevId: 299424863
-
- 06 3月, 2020 6 次提交
-
-
由 Zongwei Zhou 提交于
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
-
由 Hongkun Yu 提交于
Remove non existing samples folder mention
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 299169021
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 299160422
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 299149155
-
由 Jing Li 提交于
PiperOrigin-RevId: 299126470
-
- 05 3月, 2020 5 次提交
-
-
由 Zongwei Zhou 提交于
PiperOrigin-RevId: 299007295
-
由 Chen Chen 提交于
PiperOrigin-RevId: 298986381
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 298920502
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 298905114
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 298881095
-
- 04 3月, 2020 7 次提交
-
-
由 A. Unique TensorFlower 提交于
PiperOrigin-RevId: 298817372
-
由 Zongwei Zhou 提交于
PiperOrigin-RevId: 298734328
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 298717616
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 298692558
-
由 Yeqing Li 提交于
PiperOrigin-RevId: 298652484
-
由 Mark Daoust 提交于
-
由 Liangzhe 提交于
298416930 by lzyuan: Explicitly mark base models' state inputs as 'raw_inputs/init_lstm_h_1' and 'raw_inputs_init_lstm_h_2' when pre_bottleneck=True. -- 298380851 by skligys: Fix LSTD LSTM cells to use fixed_quantize_op(). -- 297662737 by Menglong Zhu: Explicitly replace "import tensorflow" with "tensorflow.compat.v1" for TF2.x migration -- 289667197 by lzyuan: Internal update. -- 288607438 by lzyuan: Enforce feature_extractor construction using arg keys. -- PiperOrigin-RevId: 298416930 Co-authored-by: NMenglong Zhu <menglong@google.com>
-
- 03 3月, 2020 2 次提交
-
-
由 Hongkun Yu 提交于
PiperOrigin-RevId: 298520611
-
由 Ruoxin Sang 提交于
PiperOrigin-RevId: 298493612
-