Add contributing authors to 1.4 Release notes.

Removed duplicate and removed googler from contributing author list.

Thanks!

* Disable failing Windows tests for r1.4 release.

testRemoteIteratorUsingRemoteCallOpDirectSessionGPUCPU test is failing
with "TypeError: only integer scalar arrays can be converted to a scalar
index" on the Windows GPU Release bot. Disabling test.

* Fix typo.

* Also disalbe iterator_ops_test from contrib/.

Revert "Implementing ghost batch norm as defined in https://arxiv.org…

Pin TensorBoard 0.4 to TensorFlow 1.4

This reverts commit 125f7afa.

Update version strings for TensorFlow 1.4-rc0

Update release notes for TensorFlow 1.4

Use random.sample to simplify random selection of context words

This `import` statement is now the first point where we attempt to import a generated file, and hence could see a failure if the user tries to `import tensorflow` from the root of the git repository source tree. When this `import` fails, raise a more actionable error message.

Fixes #13526.

Branch 171234659

Fix small typo in docs of learn runner.

PiperOrigin-RevId: 171234659

PiperOrigin-RevId: 171231427

PiperOrigin-RevId: 171227139

PiperOrigin-RevId: 171225190

Still missing handling control flow and pessimistic alias analysis.

PiperOrigin-RevId: 171221946

Make GCS and HDFS default build options.

PiperOrigin-RevId: 171221629

TrySimplifyAndReshapeUses should return a tensor not a node. Added a regression
test that would have failed without this CL. ArithmeticOptimizer would have
redirected the second input of concat to Split rather than Split:1.

PiperOrigin-RevId: 171220303

PiperOrigin-RevId: 171218337

If there is no exit node then there is not need to add output edges to it.

PiperOrigin-RevId: 171213900

ASAN correctly complains about the overflow on `CT(Eigen::NumTraits<U>::highest())`. This fixes the issue by providing correct CT for half and floats.

PiperOrigin-RevId: 171212745

In https://www.tensorflow.org/tutorials/wide#adding_regularization_to_prevent_overfitting, the code repeats the model_dir keyword argument, causing a syntax error if you try to run it (`SyntaxError: keyword argument repeated`).  This remove the second occurrence of the model_dir param.

* ENH: migrate ErfGrad

* TST: add test case for real value

* CLN: add semicolon

* DOC: add comment

* CLN: remove useless dependency

* CLN: remove useless dependency in LgmmaGrad

* TST: move lgamma test case

* TST: add test case for Erf

* TST: complex is unsupported for kernel

* TST: complex64 -> float

* ENH: use grad_scope

* ENH: fix grad_scope for TanhGrad and SigmoidGrad

* ENH: import M_PI

* Changed accumulate_n ==> accumulate_n_v2 and moved to contrib

* Moving source files to contrib.

* Better startup message.

* Fixing up build

* Removal of temporary code.

* Reduce logging output

* Fixing build issues.

* CI sanity fixes.

* Cleanup prior to PR

* Cleanup

* Cleanup.

* Cleanup.

* Cleanup.

* Moved AccumulateNV2 to main build and added fallback to AddN for eager mode

* Fixing CI issues

Update the Python implementation so that both Python 2 and Python 3 environment can execute

After this change when C API is enabled, function support in Python is
done with minimal use of Python code. In particular, we don't create
or store FunctionDef in Python. Small changes include:
- We don't use _hash_str for function comparisons in Python. Instead,
  we delegate this logic to TF_GraphCopyFunction in C API.
- We checking for duplication function additions from
  _DefinedFunction.add_to_graph(graph) to Graph._add_function in all
  cases. This is more logical and make it easier to support both modes.
- We change some error messages to be same in both modes.
- Since we don't store FunctionDef in C API mode in Python but get it
  on demand, access to common attributes like name or signature can
  become expensive. To mitigate this, we cache the signature (OpDef)
  of the function in Python. Signatures are generally much smaller
  than whole definitions.
- Add context manager for creating and destroying TF_Buffers.
- Allow zero output tensorflow functions in Python
  The C API and C++ runtime support functions without outputs, but Python
  APIs explicitly disallowed them before this change. This change allows
  zero output functions in Python and cleans some hacks that were added
  to side-step regular Python function APIs before.

PiperOrigin-RevId: 171201162

One is too big, three are broken due to knowon matrix_set_diag issues on windows, one is failing due to numerical discrepancies between OSs, and one is broken when multiple GPUs are present.

PiperOrigin-RevId: 171199546