Commits · 5df6a2b2b11d7aae70ee225a39eb1be25fde57ff · ModelZoo / ResNet50_tensorflow

30 Apr, 2021 2 commits
- Internal change · 5df6a2b2
  Abdullah Rashwan authored Apr 30, 2021
```
PiperOrigin-RevId: 371418693
```
  5df6a2b2
- Internal change · fa4cb013
  Abdullah Rashwan authored Apr 30, 2021
```
PiperOrigin-RevId: 371389966
```
  fa4cb013
26 Apr, 2021 1 commit
- Add adagrad optimizer support. · 834c0454
  Hao Wu authored Apr 26, 2021
```
PiperOrigin-RevId: 370551493
```
  834c0454
21 Apr, 2021 1 commit
- Update `Config` class doc to warn the auto conversion of dict to `Config` class. · bcc4f9d6
  Hongkun Yu authored Apr 20, 2021
```
PiperOrigin-RevId: 369588823
```
  bcc4f9d6
20 Apr, 2021 1 commit
- Expose learning rate schedule classes. · b51b9342
  Hongkun Yu authored Apr 19, 2021
```
PiperOrigin-RevId: 369375299
```
  b51b9342
14 Apr, 2021 2 commits
- Internal change · 68b8a50d
  Hongkun Yu authored Apr 14, 2021
```
PiperOrigin-RevId: 368535232
```
  68b8a50d
- Change activation/add to a keras layer form for the ResNet50, MobilenetV2 default case. · 9cc08b70
  Jaehong Kim authored Apr 14, 2021
```
This CL changes ResidualBlock and InvertedBottleneckBlock.

PiperOrigin-RevId: 368383954
```
  9cc08b70
12 Apr, 2021 1 commit

Use nonexperimental mixed precision API for official models. · ba8ad4f5

Reed Wanderman-Milne authored Apr 12, 2021

For all modified calls to set_mixed_precision_policy(), the loss_scale argument was removed, as it cannot be passed if the nonexperimental API is used. For all such callers, the loss_scale is later used to explicitly create a LossScaleOptimizer, so removing the argument has no impact.

Switching to the non-experimental LossScaleOptimizer has no effect, as it has near identical behavior and all isinstance checks within the official models check for the non-experimental version.

PiperOrigin-RevId: 368101975

ba8ad4f5

06 Apr, 2021 3 commits

Change default value of use_experimental_api to False. · 75b52393
Reed Wanderman-Milne authored Apr 06, 2021
```
PiperOrigin-RevId: 367105004
```
75b52393

Pass use_experimental_api=True to mixed precision functions. · ef8f9c26

Reed Wanderman-Milne authored Apr 06, 2021

This has no functional impact since the default is currently True, but I plan on changing the default to False soon.

PiperOrigin-RevId: 367049942

ef8f9c26

Use tf.compat.v1 version of enable_mixed_precision_graph_rewrite. · 75d9b66f

Reed Wanderman-Milne authored Apr 06, 2021

The function `tf.train.experimental.enable_mixed_precision_graph_rewrite` will be removed from the TF2 namespace soon, at which point it will only be accessible under tf.compat.v1.

PiperOrigin-RevId: 367046393

75d9b66f

30 Mar, 2021 1 commit
- Internal change · cc192f26
  A. Unique TensorFlower authored Mar 29, 2021
```
PiperOrigin-RevId: 365713370
```
  cc192f26
25 Mar, 2021 1 commit
- Open source the second half of multi-task library · e773b9b3
  A. Unique TensorFlower authored Mar 25, 2021
```
PiperOrigin-RevId: 365085378
```
  e773b9b3
18 Mar, 2021 2 commits
- [core] Add tpu_enable_xla_dynamic_padder to RuntimeConfig. · 64d412cb
  Frederick Liu authored Mar 18, 2021
```
PiperOrigin-RevId: 363595377
```
  64d412cb
- [core] Add tpu_enable_xla_dynamic_padder to RuntimeConfig. · 124e3918
  Frederick Liu authored Mar 18, 2021
```
PiperOrigin-RevId: 363595377
```
  124e3918
06 Mar, 2021 2 commits
- Clean up unnecessary code in PY3 · 9e46a6cb
  Hongkun Yu authored Mar 06, 2021
```
PiperOrigin-RevId: 361348924
```
  9e46a6cb
- Clean up unnecessary code in PY3 · 3e45d52f
  Hongkun Yu authored Mar 06, 2021
```
PiperOrigin-RevId: 361348924
```
  3e45d52f
03 Mar, 2021 4 commits
- Pass use_experimental_api=True to mixed precision functions. · 5b9f6e2f
  Reed Wanderman-Milne authored Mar 03, 2021
```
The default is True, but I plan on changing it to False soon. After that, I plan on removing the argument and never using the experimental API.

PiperOrigin-RevId: 360724698
```
  5b9f6e2f
- Pass use_experimental_api=True to mixed precision functions. · 4ab6f1b1
  Reed Wanderman-Milne authored Mar 03, 2021
```
The default is True, but I plan on changing it to False soon. After that, I plan on removing the argument and never using the experimental API.

PiperOrigin-RevId: 360724698
```
  4ab6f1b1
- Fix corner cases where LR schedules output Inf. · c9ab5a7a
  Le Hou authored Mar 03, 2021
```
PiperOrigin-RevId: 360719881
```
  c9ab5a7a
- Fix corner cases where LR schedules output Inf. · 067d3e41
  Le Hou authored Mar 03, 2021
```
PiperOrigin-RevId: 360719881
```
  067d3e41
02 Mar, 2021 2 commits
- Internal change · c4ac240d
  Le Hou authored Mar 02, 2021
```
PiperOrigin-RevId: 360489888
```
  c4ac240d
- Internal change · 249dcac3
  Le Hou authored Mar 02, 2021
```
PiperOrigin-RevId: 360489888
```
  249dcac3
01 Mar, 2021 2 commits
- Add global_step to reduce_aggregated_logs(). · 7ea62b0b
  A. Unique TensorFlower authored Mar 01, 2021
```
PiperOrigin-RevId: 360256877
```
  7ea62b0b
- Add global_step to reduce_aggregated_logs(). · 48192d54
  A. Unique TensorFlower authored Mar 01, 2021
```
PiperOrigin-RevId: 360256877
```
  48192d54
26 Feb, 2021 2 commits
- Internal change · ff52d0b3
  Le Hou authored Feb 26, 2021
```
PiperOrigin-RevId: 359821383
```
  ff52d0b3
- Internal change · cd0f708a
  Le Hou authored Feb 26, 2021
```
PiperOrigin-RevId: 359821383
```
  cd0f708a
17 Feb, 2021 2 commits

[core] Use the full model instead of checkpoint_items when constructing... · c013311e

Frederick Liu authored Feb 17, 2021

[core] Use the full model instead of checkpoint_items when constructing Checkpoint in multitask evaluator.

PiperOrigin-RevId: 357962526

c013311e

[core] Use the full model instead of checkpoint_items when constructing... · ad66ddb4

Frederick Liu authored Feb 17, 2021

[core] Use the full model instead of checkpoint_items when constructing Checkpoint in multitask evaluator.

PiperOrigin-RevId: 357962526

ad66ddb4

15 Feb, 2021 2 commits

[core] Update BestCheckpointExporter to support dictionary with arbitrary... · 1ef32d1a

Frederick Liu authored Feb 15, 2021

[core] Update BestCheckpointExporter to support dictionary with arbitrary depth (possibly coming from multitask eval).
Also added best checkpointexporter support to multitask.

PiperOrigin-RevId: 357584596

1ef32d1a

[core] Update BestCheckpointExporter to support dictionary with arbitrary... · 84a561ca

Frederick Liu authored Feb 15, 2021

[core] Update BestCheckpointExporter to support dictionary with arbitrary depth (possibly coming from multitask eval).
Also added best checkpointexporter support to multitask.

PiperOrigin-RevId: 357584596

84a561ca

07 Feb, 2021 2 commits
- Internal change · bde9cdca
  Le Hou authored Feb 07, 2021
```
PiperOrigin-RevId: 356158181
```
  bde9cdca
- Internal change · 5fa7fbea
  Le Hou authored Feb 07, 2021
```
PiperOrigin-RevId: 356158181
```
  5fa7fbea
22 Jan, 2021 7 commits
- Internal change · a2201bb8
  Le Hou authored Jan 22, 2021
```
PiperOrigin-RevId: 353327814
```
  a2201bb8
- Internal change · 39abb95d
  Le Hou authored Jan 22, 2021
```
PiperOrigin-RevId: 353327814
```
  39abb95d
- Respectful code cleanup · 12db3d34
  Hongkun Yu authored Jan 22, 2021
```
PiperOrigin-RevId: 353315689
```
  12db3d34
- Respectful code cleanup · cc9357f9
  Hongkun Yu authored Jan 22, 2021
```
PiperOrigin-RevId: 353315689
```
  cc9357f9
- Fix typo. Found by https://github.com/tensorflow/models/issues/9656 · b0645a5c
  Hongkun Yu authored Jan 22, 2021
```
PiperOrigin-RevId: 353277457
```
  b0645a5c
- Fix typo. Found by https://github.com/tensorflow/models/issues/9656 · 9f3f80e2
  Hongkun Yu authored Jan 22, 2021
```
PiperOrigin-RevId: 353277457
```
  9f3f80e2
- Internal change · 5e8a51fd
  Steven Hickson authored Jan 22, 2021
```
PiperOrigin-RevId: 353260280
```
  5e8a51fd