GitLab

Add text summarization model to tensorflow/models.

* Fix bug in inception README.md

Fixes #59

* Document that number of classes and example count is hardcoded

Fixes #190

fixed URL for pretrained syntactic models

* Adding scripts for tokenization and parsing in new languages.

* Adding:
  - offset feature locator,
  - last word feature function,
  - untokenized text format.

* Adding references to @calberti and @andorardo to syntaxnet readme.

* Morpher and segmenter transition systems and new features (quotes, punctuation, capitalization, character ngrams, morphology attributes).

Cleanup changes for syntaxnet.

Fixed confusing wording

The script actually reads $LABEL_FILE and strips out the text, which it uses to label and find image files in their respective sub directories.

* TF defines the same external repositories but with different commit and BUILD

* Fix a bunch of bazel namespace issues

@tf changed to @org_tensorflow
Tensorflow stopped using protobuf as submodule, now it's external, so use @protobuf

* Sync to master of tensorflow submodule:
  - local version of is_cuda in syntaxnet.bzl,
  - using unique_ptr in file api.

As Docker builds are layer by layer, removing cached package files in different layers do not reduce image size. To reduce image size, cached packages have to removed in the same layer i.e., `RUN` command as the packages have been installed.
Signed-off-by: Riyad Parvez <social.riyad@gmail.com>

Misspelling in parameter (factfor --> factor)

* Use the tensorflow cross entropy function to prevent nan losses.

* Correct double softmax and use mean for loss.

remove unneeded assignment statement

Fix linkage of UnpackSparseFeatures op so kernel is registered

Error Fix in Read.me, Added download_and_preprocess_flowers_mac for Mac commands

fix for broken link to slim README.md

Update slim README.md

Fixed references to slim.losses

Update neural_gpu documentation.

* Switch from decimal to symbol in most places and add a description of what a symbol is per Lukasz.

* Add list of algorithmic tasks.
* Detail what is stored in /tmp/neural_gpu/ (default location).
* Detail that one should hit Ctrl-C to stop the training process.
* Add sections around decoding and animation, but couldn't get those to work myself. Still worth having in the docs.

fixes b/28821889

When importing tensorflow before scipi.ndimage and running the
example in a docker container, the following error is given:

```
 % python example.py
Traceback (most recent call last):
  File "example.py", line 25, in <module>
    im = im / 255.
TypeError: unsupported operand type(s) for /: 'instance' and 'float'
```
This only happens for 'cat.jpg'. When converting the image to a '.png'
file, the example runs as expected.

When swapping the imports around so that scipy.ndimage is imported
BEFORE tensorflow, both 'jpg' and 'png' files work.