Commits · 9485aa1dcf2c2e1ebaede0af0283796dcfb01911 · ModelZoo / ResNet50_tensorflow

10 Mar, 2021 2 commits
- Internal change · 5981350b
  Frederick Liu authored Mar 09, 2021
```
PiperOrigin-RevId: 361957289
```
  5981350b
- Internal change · 5a898973
  Frederick Liu authored Mar 09, 2021
```
PiperOrigin-RevId: 361957289
```
  5a898973
26 Oct, 2020 2 commits
- Py3 cleanup for nlp/ · cd5e1204
  Hongkun Yu authored Oct 26, 2020
```
PiperOrigin-RevId: 339071563
```
  cd5e1204
- Py3 cleanup for nlp/ · 89fd2bb1
  Hongkun Yu authored Oct 26, 2020
```
PiperOrigin-RevId: 339071563
```
  89fd2bb1
06 Sep, 2020 2 commits

Fix a bug that the transformer model cannot be saved when save_weights_only is... · fe46c155

Chen Chen authored Sep 06, 2020

Fix a bug that the transformer model cannot be saved when save_weights_only is set to False: https://github.com/tensorflow/models/issues/9186

PiperOrigin-RevId: 330292460

fe46c155

Fix a bug that the transformer model cannot be saved when save_weights_only is... · c60951b1

Chen Chen authored Sep 06, 2020

Fix a bug that the transformer model cannot be saved when save_weights_only is set to False: https://github.com/tensorflow/models/issues/9186

PiperOrigin-RevId: 330292460

c60951b1

12 Aug, 2020 2 commits
- Internal change · 999fae62
  Hongkun Yu authored Aug 12, 2020
```
PiperOrigin-RevId: 326286926
```
  999fae62
- Internal change · 88253ce5
  Hongkun Yu authored Aug 12, 2020
```
PiperOrigin-RevId: 326286926
```
  88253ce5
17 Jun, 2020 1 commit
- Internal change · 6da061c0
  Abdullah Rashwan authored Jun 16, 2020
```
PiperOrigin-RevId: 316797555
```
  6da061c0
24 May, 2020 1 commit
- [Clean up]: remove is_v2() check inside transformer. · b29fe6b7
  Hongkun Yu authored May 24, 2020
```
PiperOrigin-RevId: 312988874
```
  b29fe6b7
14 May, 2020 1 commit
- Remove profiler callback which is calling tensorflow internal library. · ec3a4616
  Hongkun Yu authored May 13, 2020
```
Profiler has been integrated with TensorBoard callback and released in TF 2.2

PiperOrigin-RevId: 311475711
```
  ec3a4616
06 May, 2020 1 commit
- Update steps_between_evals to make it does not break initially · 34dd50a3
  Hongkun Yu authored May 06, 2020
```
PiperOrigin-RevId: 310228606
```
  34dd50a3
27 Apr, 2020 1 commit
- Test Transformer training and evaluation with MLIR TF to XLA bridge enabled. · 0886b384
  Lucy Fox authored Apr 27, 2020
```
PiperOrigin-RevId: 308666667
```
  0886b384
22 Apr, 2020 1 commit
- Internal change · c1468202
  Tayo Oguntebi authored Apr 22, 2020
```
PiperOrigin-RevId: 307918408
```
  c1468202
25 Mar, 2020 1 commit
- Internal change · 7c83a9d7
  Will Cromar authored Mar 25, 2020
```
PiperOrigin-RevId: 302921283
```
  7c83a9d7
10 Mar, 2020 1 commit
- Internal change · 08f45dc4
  A. Unique TensorFlower authored Mar 10, 2020
```
PiperOrigin-RevId: 300203487
```
  08f45dc4
05 Mar, 2020 1 commit
- Remove force_v2_in_keras_compile. experimental_run_tf_function is no-op now. · d3d7f15f
  Hongkun Yu authored Mar 05, 2020
```
PiperOrigin-RevId: 299160422
```
  d3d7f15f
25 Feb, 2020 1 commit

Convert numpy arrays to floats using float constructor, not .item(). · ed2e3bc3

Reed Wanderman-Milne authored Feb 25, 2020

I plan on changing Keras to return Python floats instead of Numpy scalars, and this will break if .item() is called. Also, using the float constructor is cleaner, as it clearly communicates that a Python float is desired.

PiperOrigin-RevId: 297192289

ed2e3bc3

13 Feb, 2020 1 commit
- [Refactor] TF models: move all contents of transformer to nlp/transformer · 7f926353
  Hongkun Yu authored Feb 13, 2020
```
PiperOrigin-RevId: 294997928
```
  7f926353
19 Dec, 2019 1 commit
- Refactor: move transformer/model to nlp/transformer. · 1722b691
  Hongkun Yu authored Dec 18, 2019
```
PiperOrigin-RevId: 286325224
```
  1722b691
12 Dec, 2019 1 commit
- Enabled performance related parameters for Transformer: all_reduce_alg, · 152adcbd
  A. Unique TensorFlower authored Dec 12, 2019
```
enable_eager, tf_gpu_thread_mode and datasets_num_private_threads

PiperOrigin-RevId: 285157129
```
  152adcbd
27 Nov, 2019 2 commits
- Enabled performance related parameters for Transformer: all_reduce_alg, · bd73fdfe
  Hongkun Yu authored Nov 27, 2019
```
enable_eager, tf_gpu_thread_mode and datasets_num_private_threads

PiperOrigin-RevId: 282804724
```
  bd73fdfe
- Enabled performance related parameters for Transformer: all_reduce_alg, · 18df1a09
  A. Unique TensorFlower authored Nov 27, 2019
```
enable_eager, tf_gpu_thread_mode and datasets_num_private_threads

PiperOrigin-RevId: 282797514
```
  18df1a09
25 Nov, 2019 1 commit
- Internal change · 5b2575c2
  Zongwei Zhou authored Nov 24, 2019
```
PiperOrigin-RevId: 282293668
```
  5b2575c2
24 Oct, 2019 1 commit
- Remove unnecessary flags · cd00b9a7
  Hongkun Yu authored Oct 24, 2019
```
PiperOrigin-RevId: 276518206
```
  cd00b9a7
09 Sep, 2019 1 commit

Unexpose some flags from models which do not use them. · e91c41c2

Reed Wanderman-Milne authored Sep 09, 2019

--stop_threshold, --num_gpu, --hooks, --export_dir, and --distribution_strategy have been unexposed from models which do not use them

PiperOrigin-RevId: 268032080

e91c41c2

04 Sep, 2019 1 commit

Unexpose some flags from models which do not use them. · a85c40e3

Reed Wanderman-Milne authored Sep 03, 2019

--clean, --train_epochs, and --epochs_between_evals have been unexposed from models which do not use them

PiperOrigin-RevId: 267065651

a85c40e3

30 Aug, 2019 1 commit
- Internal change · e170a8ba
  A. Unique TensorFlower authored Aug 30, 2019
```
PiperOrigin-RevId: 266413847
```
  e170a8ba
22 Aug, 2019 1 commit
- Internal change · 640ff472
  A. Unique TensorFlower authored Aug 22, 2019
```
PiperOrigin-RevId: 264853703
```
  640ff472
20 Aug, 2019 1 commit
- fix transformer amp · c9c05e9b
  Vinh Nguyen authored Aug 20, 2019
  
  c9c05e9b
16 Aug, 2019 1 commit
- Consolidation & readability. · b1d9ac5b
  Hongkun Yu authored Aug 16, 2019
```
PiperOrigin-RevId: 263863438
```
  b1d9ac5b
15 Aug, 2019 1 commit
- Internal change · 47e783e6
  A. Unique TensorFlower authored Aug 14, 2019
```
PiperOrigin-RevId: 263485329
```
  47e783e6
09 Aug, 2019 1 commit
- adding automatic mixed precision training to transformer v2 · b8f58ec8
  Vinh Nguyen authored Aug 09, 2019
  
  b8f58ec8
23 Jul, 2019 1 commit

Single execution path tests for ResNet50, ResNet56, NCF, and Shakespeare LSTM. (#7276) · 9d8c9aa4

Toby Boyd authored Jul 23, 2019

* Add force_run_distributed tests.

* Added enable_eager

* r/force_run_distributed/force_v2_in_keras_compile

* Adding force_v2 tests and FLAGs.

* Rename method to avoid conflict.

* Add cpu force_v2 tests.

* fix lint, wrap line.

* change to force_v2_in_keras_compile

* Update method name.

* Lower mlperf target to 0.736.

9d8c9aa4

21 Jun, 2019 1 commit
- Transformer 2.0: Make metrics layer optional (#7075) · 513fdbb2
  guptapriya authored Jun 21, 2019
```
* trying fake merge call

* make metrics optional

* Remove extra print
```
  513fdbb2
19 Jun, 2019 2 commits

Add mixed precision support to Transformer (#7011) · f8ec01ae
Reed authored Jun 19, 2019

f8ec01ae

Add XLA to transformer (#7048) · 269581dc

Toby Boyd authored Jun 19, 2019



* set default steps to 300K.

* Log flags to perfzero.

* Add XLA support to transformer

- Moved config logic to keras_utils
- Added enable_xla flag to _performance flags
- Did not refactor enable_xla flag from keras resnet due to
  reliance on calling FLAGs in estimator keras and that is
  a needed refactor for another time.

* fix g3 lint complaint.

* Refactor set config into keras_utils.

* Move flags out of main.

* pipe through enable_xla

* Update official/transformer/v2/misc.py
Co-Authored-By: Reed <reedwm@google.com>

269581dc

18 Jun, 2019 1 commit
- set default steps to 300K. (#7046) · 01c3a4f3
  Toby Boyd authored Jun 18, 2019
  
  01c3a4f3
29 May, 2019 1 commit

Make max_length and static_batch configurable (#6893) · ab1c1dfc

Zhang Xunkai authored May 29, 2019

* Make max_length and static_batch configurable.

* Fix line length.

* Fix incorrect parameters in building eval input.

* Improve comments for readability.

ab1c1dfc

24 May, 2019 1 commit

Transformer v2 benchmark (#6860) · f2ea2f53

Toby Boyd authored May 24, 2019

* Moved common keras code to utils.

* Initial 1 gpu benchmark

- Aligned flags with resnet example
- removed code/features that are not super useful
- eval as part of train if bleu source/ref provided
- add exp_per_second hook

* Rename benchmark classes, pass batch-size and log_steps.

* fix docstring

* Predict done with checkpoints inline

- perfzero baseclass

* steps not epochs with smoother training loop.

* do not initialize history outside loop.

* 5000 between eval not 500

* estimator to keras.

* remove epochs var.

* use range not xrange.

* 200K steps for 1 gpu

* fix global step

f2ea2f53