Commits · f7f2bf4f416addb6b5c1f7df67bfff13eb48ddd5 · ModelZoo / ResNet50_tensorflow

17 Apr, 2018 6 commits
- Fix TensorRT test output and add int8 result. (#4004) · f7f2bf4f
  gracehoney authored Apr 17, 2018
```
* Fix TensorRT test output and add int8 result.

* Fix comments

* Removing footnote declaration as well
```
  f7f2bf4f
- Merge pull request #3998 from alasin/master · e8863ed6
  aquariusjay authored Apr 17, 2018
```
Typo in ADE20K instructions.
```
  e8863ed6
- Bugfix for initialization in autoencoder. (#3477) · 3f7d667e
  Swaroop Guntupalli authored Apr 17, 2018
  
  3f7d667e
- Typo in ADE20K instructions. · db5850b9
  Anuj Pahuja authored Apr 17, 2018
  
  db5850b9
- Merge pull request #3973 from pkulzc/master · ed4e22b8
  pkulzc authored Apr 16, 2018
```
Object detection internal changes
```
  ed4e22b8
- Sync to master. · 13b89b93
  pkulzc authored Apr 16, 2018
  
  13b89b93
16 Apr, 2018 3 commits
- Merge pull request #3977 from tensorflow/MarkDaoust-patch-2 · cac90a0e
  Neal Wu authored Apr 16, 2018
```
Switch image link to https. Devsite wont load this over http.
```
  cac90a0e
- Fix copyright year. · a4b6765a
  pkulzc authored Apr 16, 2018
  
  a4b6765a
- fix issue 'could not satisfy explicit device' (#3641) · c7c3970c
  Shaoning Zeng authored Apr 17, 2018
```
* fix issue 'could not satisfy explicit device'

* remove the line unrelated to fix issue
```
  c7c3970c
13 Apr, 2018 23 commits
- Switch image link to https. Devsite wont load this over http. · 490813bd
  Mark Daoust authored Apr 13, 2018
  
  490813bd
- Update to trainer to allow for reading multiclass scores · cc08dc87
  Zhichao Lu authored Apr 12, 2018
```
PiperOrigin-RevId: 192624207
```
  cc08dc87
- Contains implementation of Visual Relations Detection evaluation metric (per · 1bddd18e
  Zhichao Lu authored Apr 12, 2018
```
image evaluation).

PiperOrigin-RevId: 192583425
```
  1bddd18e
- Modify the ssd meta arch to allow the option of not adding an implicit background class. · eccae449
  Zhichao Lu authored Apr 11, 2018
```
PiperOrigin-RevId: 192529600
```
  eccae449
- Refactor model_tpu_main.py files and move continuous eval loop into model_lib.py · a60dd985
  Zhichao Lu authored Apr 11, 2018
```
PiperOrigin-RevId: 192512429
```
  a60dd985
- Update create_pascal_tf_record.py and create_pet_tf_record.py · f98f000a
  Zhichao Lu authored Apr 11, 2018
```
PiperOrigin-RevId: 192494267
```
  f98f000a
- Enforcing that all eval metric ops have valid python strings. · 2151c447
  Zhichao Lu authored Apr 11, 2018
```
PiperOrigin-RevId: 192485456
```
  2151c447
- Set regularize_depthwise to true in mobilenet_v1_argscope. · 0940134e
  Zhichao Lu authored Apr 11, 2018
```
PiperOrigin-RevId: 192472546
```
  0940134e
- Refactoring of Mask-RCNN to put all mask prediction code in third stage. · b47ca971
  Zhichao Lu authored Apr 11, 2018
```
PiperOrigin-RevId: 192421843
```
  b47ca971
- Returning eval_on_train_input_fn from create_estimator_and_inputs(), rather... · 227f41e9
  Zhichao Lu authored Apr 10, 2018
```
Returning eval_on_train_input_fn from create_estimator_and_inputs(), rather than using train_input_fn in EVAL mode (which will still have data augmentation).

PiperOrigin-RevId: 192320460
```
  227f41e9
- Access TPUEstimator and CrossShardOptimizer from tf namesspace. · 7e810001
  Zhichao Lu authored Apr 09, 2018
```
PiperOrigin-RevId: 192226678
```
  7e810001
- Fix test that was flaky due to randomness · b0c5c3b5
  Zhichao Lu authored Apr 09, 2018
```
PiperOrigin-RevId: 192195514
```
  b0c5c3b5
- Minor fixes to match git repo. · 2624a23e
  Zhichao Lu authored Apr 09, 2018
```
PiperOrigin-RevId: 192166224
```
  2624a23e
- use shape utils for assertion in feature extractor. · e25c014e
  Zhichao Lu authored Apr 09, 2018
```
PiperOrigin-RevId: 192147130
```
  e25c014e
- Class agnostic masks for mask_rcnn · a4d9c3a0
  Zhichao Lu authored Apr 09, 2018
```
PiperOrigin-RevId: 192132440
```
  a4d9c3a0
- Add learning rate summary in EVAL mode in model.py · bfd15ec1
  Zhichao Lu authored Apr 07, 2018
```
PiperOrigin-RevId: 192006190
```
  bfd15ec1
- Migrating away from Experiment class, as it is now deprecated. Also,... · 90cc9baa
  Zhichao Lu authored Apr 07, 2018
```
Migrating away from Experiment class, as it is now deprecated. Also, refactoring into a separate model library and binaries.

PiperOrigin-RevId: 192004845
```
  90cc9baa
- Add classification_loss and localiztion_loss metrics for TPU jobs. · 8deba73f
  Zhichao Lu authored Apr 06, 2018
```
PiperOrigin-RevId: 191957195
```
  8deba73f
- Add an option to skip the last striding in mobilenet. The modified network has... · aedfa2e4
  Zhichao Lu authored Apr 06, 2018
```
Add an option to skip the last striding in mobilenet. The modified network has nominal output stride 16 instead of 32.

PiperOrigin-RevId: 191932855
```
  aedfa2e4
- Add option to override base feature extractor hyperparams in SSD models. This... · decbad8a
  Zhichao Lu authored Apr 05, 2018
```
Add option to override base feature extractor hyperparams in SSD models. This would allow us to use the same set of hyperparams for the complete feature extractor (base + new layers) if desired.

PiperOrigin-RevId: 191787921
```
  decbad8a
- Adding an attribute to SSD model to indicate which fields in prediction... · 45ecc0f9
  Zhichao Lu authored Apr 05, 2018
```
Adding an attribute to SSD model to indicate which fields in prediction dictionary have a batch dimension. This will be useful for future video models.

PiperOrigin-RevId: 191743097
```
  45ecc0f9
- Internal change. · a0c3c440
  Zhichao Lu authored Apr 04, 2018
```
PiperOrigin-RevId: 191668425
```
  a0c3c440
- Sync to latest master. · b668f594
  pkulzc authored Apr 13, 2018
  
  b668f594
12 Apr, 2018 4 commits

Resnet distribution strategies (#3887) · 32aa6563

Taylor Robie authored Apr 12, 2018

* begin transfer from contrib fork

more changes to resnet_run_loop

use AUTOTUNE in prefetch

first pass at resnet with functional distribution strategies

fix syntax error

delint

aesthetic tweaks

delint and fix typos

rip multi_gpu flag out of resnet entirely. Subject to saved model load verification

update cifar10 and imagenet tests to reflect that the model function no longer need to know about multi_gpu

fix imagenet test

start addressing PR comments

more PR response work

* misc tweaks

* add a comment

* final pr tweaks

* fix parsers

32aa6563

Add link to the paper and the Tratz dataset (#3745) · ad7755c8
Vered Shwartz authored Apr 12, 2018

ad7755c8
Merge pull request #3957 from franckverrot/patch-1 · f20fac82
Mark Daoust authored Apr 11, 2018
```
Correct a small typo
```
f20fac82
Correct a small typo · b87f5a07
Franck Verrot authored Apr 11, 2018
```
This commit changes a tiny typo in the Eager Execution notebook.
```
b87f5a07

11 Apr, 2018 4 commits

Merge pull request #3942 from asimshankar/mnist_sequential_and_estimator · 2c74e416
Asim Shankar authored Apr 11, 2018
```
official/mnist: Use tf.keras.Sequential
```
2c74e416

Fix order of variable update and moving average update · 720d3363

Igor Ganichev authored Apr 11, 2018

While the original code works fine in practice, it technically
allows gradient application and moving average update to happen
in any order. This causes the behavior to deviate from pure
mathematical specifications.

720d3363

Fix order of variable update and moving average update · f6e4dfe9

Igor Ganichev authored Apr 10, 2018

While the original code works fine in practice, it technically
allows gradient application and moving average update to happen
in any order. This causes the behavior to deviate from pure
mathematical specifications.

f6e4dfe9

official/mnist: Address comment in PR · cfe945ef
Asim Shankar authored Apr 10, 2018

cfe945ef