GitLab

This commit changes a tiny typo in the Eager Execution notebook.

official/mnist: Use tf.keras.Sequential

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.

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.

* change reference_data.py to use tf.gfile

* simplify json treatment

* Update reference files to account for a superficial change in batch_norm

Fix string to bytes conversion in deeplab data input. #3885

The current code does not work in py2 since the input string could
contain unicode string, and default encoding in ASCII in py2. Change
the method to only do encode() which convert string to byte array
when running python3.

* Adding tests

* Adding tests

* Repackaging

* Adding logging

* Linting

Updating DELF README

Update create_pascal_tf_record.py

Update evaluation_protocols.md

* Add fp16 support to resnet.

* address PR comments

* add dtype checking to model definition

* delint

* more PR comments

* few more tweaks

* update resnet checkpoints

add ADE20K dataset

* Adding cast to fp32

* Adding myself to OWNERS

Update evaluation_protocols.md

duplicate eval metric: change the fourth metric from ## Weighted PASCAL VOC detection metric to ##  Weighted PASCAL VOC instance segmentation metric .

fix typo

Touch-ups to eager's getting started guide

Fix typo

* "Machine Learning" --> Machine learning.
* "steepest the ascent" --> "steepest ascent"
* "standard gradient descent" --> "stochastic gradient descent"

Add cyclegan to open source tensorflow/models

Add input images, labels, and output semantic predictions to summary