Cherrypicks

RELEASE.md

@@ -3,9 +3,8 @@
 ## Major Features and Improvements
 * Added `tf.layers.conv3d_transpose` layer for spatio temporal deconvolution.
 * Added `tf.Session.make_callable()`, which provides a lower overhead means of running a similar step multiple times.
-* Added libverbs-based RDMA support to contrib (courtesy @junshi15 from Yahoo).
-* Bring `tf.feature_column.*` into the API. Non-deprecated functionality from `tf.contrib.layers.*` is moved to `tf.feature_column.*` with cosmetic changes.
-* `RNNCell` objects now subclass `tf.layers._Layer`.  The strictness described
+* Added ibverbs-based RDMA support to contrib (courtesy @junshi15 from Yahoo).
+* `RNNCell` objects now subclass `tf.layers.Layer`.  The strictness described
   in the TensorFlow 1.1 release is gone:  The first time an RNNCell is used,
   it caches its scope.  All future uses of the RNNCell will reuse variables from
   that same scope.  This is a breaking change from the behavior of RNNCells
@@ -23,6 +22,28 @@
 * TensorFlow C library now available for Windows.
 * We released a new open-source version of TensorBoard.
 * [`SavedModel CLI`](https://www.tensorflow.org/versions/master/programmers_guide/saved_model_cli) tool available to inspect and execute MetaGraph in SavedModel
+* RNNCells' variable names have been renamed for consistency with Keras layers.
+  Specifically, the previous variable names "weights" and "biases" have
+  been changed to "kernel" and "bias", respectively.
+  This may cause backward incompatibility with regard to your old
+  checkpoints containing such RNN cells, in which case you can use the tool
+  [checkpoint_convert script](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/contrib/rnn/python/tools/checkpoint_convert.py)
+  to convert the variable names in your old checkpoints.
+* Many of the RNN functions and classes that were in the `tf.nn` namespace
+  before the 1.0 release and which were moved to `tf.contrib.rnn` have now
+  been moved back to the core namespace.  This includes
+  `RNNCell`, `LSTMCell`, `GRUCell`, and a number of other cells.  These
+  now reside in `tf.nn.rnn_cell` (with aliases in `tf.contrib.rnn` for backwards
+  compatibility).  The original `tf.nn.rnn` function is now `tf.nn.static_rnn`,
+  and the bidirectional static and state saving static rnn functions are also
+  now back in the `tf.nn` namespace.
+
+  Notable exceptions are the `EmbeddingWrapper`, `InputProjectionWrapper` and
+  `OutputProjectionWrapper`,  which will slowly be moved to deprecation
+  in `tf.contrib.rnn`.  These are inefficient wrappers that should often
+  be replaced by calling `embedding_lookup` or `layers.dense` as pre- or post-
+  processing of the rnn.  For RNN decoding, this functionality has been replaced
+  with an alternative API in `tf.contrib.seq2seq`.
 
 ## Breaking Changes to the API
 * `org.tensorflow.contrib.android.TensorFlowInferenceInterface` now throws exceptions where possible and has simplified method signatures.

tensorflow/c/c_api.h

@@ -961,8 +961,9 @@ typedef struct TF_WhileParams {
 // - Reference-type inputs
 // - Directly referencing external tensors from the cond/body graphs (this is
 //   possible in the Python API)
-TF_WhileParams TF_NewWhile(TF_Graph* g, TF_Output* inputs, int ninputs,
-                           TF_Status* status);
+TF_CAPI_EXPORT extern TF_WhileParams TF_NewWhile(TF_Graph* g, TF_Output* inputs,
+                                                 int ninputs,
+                                                 TF_Status* status);
 
 // Builds the while loop specified by `params` and returns the output tensors of
 // the while loop in `outputs`. `outputs` should be allocated to size
@@ -972,13 +973,14 @@ TF_WhileParams TF_NewWhile(TF_Graph* g, TF_Output* inputs, int ninputs,
 //
 // Either this or TF_AbortWhile() must be called after a successful
 // TF_NewWhile() call.
-void TF_FinishWhile(const TF_WhileParams* params, TF_Status* status,
-                    TF_Output* outputs);
+TF_CAPI_EXPORT extern void TF_FinishWhile(const TF_WhileParams* params,
+                                          TF_Status* status,
+                                          TF_Output* outputs);
 
 // Frees `params`s resources without building a while loop. `params` is no
 // longer valid after this returns. Either this or TF_FinishWhile() must be
 // called after a successful TF_NewWhile() call.
-void TF_AbortWhile(const TF_WhileParams* params);
+TF_CAPI_EXPORT extern void TF_AbortWhile(const TF_WhileParams* params);
 
 // Adds operations to compute the partial derivatives of sum of `y`s w.r.t `x`s,
 // i.e., d(y_1 + y_2 + ...)/dx_1, d(y_1 + y_2 + ...)/dx_2...
@@ -994,8 +996,9 @@ void TF_AbortWhile(const TF_WhileParams* params);
 // supports. See
 // https://www.tensorflow.org/code/tensorflow/cc/gradients/README.md
 // for instructions on how to add C++ more gradients.
-void TF_AddGradients(TF_Graph* g, TF_Output* y, int ny, TF_Output* x, int nx,
-                     TF_Output* dx, TF_Status* status, TF_Output* dy);
+TF_CAPI_EXPORT void TF_AddGradients(TF_Graph* g, TF_Output* y, int ny,
+                                    TF_Output* x, int nx, TF_Output* dx,
+                                    TF_Status* status, TF_Output* dy);
 
 // TODO(josh11b): Register OpDef, available to all operations added
 // to this graph.
@@ -1032,7 +1035,7 @@ TF_CAPI_EXPORT extern TF_Session* TF_NewSession(TF_Graph* graph,
 //
 // If successful, populates `graph` with the contents of the Graph and
 // `meta_graph_def` with the MetaGraphDef of the loaded model.
-TF_Session* TF_LoadSessionFromSavedModel(
+TF_CAPI_EXPORT extern TF_Session* TF_LoadSessionFromSavedModel(
     const TF_SessionOptions* session_options, const TF_Buffer* run_options,
     const char* export_dir, const char* const* tags, int tags_len,
     TF_Graph* graph, TF_Buffer* meta_graph_def, TF_Status* status);

tensorflow/contrib/android/README.md

@@ -27,10 +27,17 @@ allprojects {
 }
 
 dependencies {
-    compile 'org.tensorflow:tensorflow-android:1.2.0-preview'
+    compile 'org.tensorflow:tensorflow-android:+'
 }
 ```
 
+This will tell Gradle to use the
+[latest version](https://bintray.com/google/tensorflow/tensorflow-android/_latestVersion)
+of the TensorFlow AAR that has been released to
+[https://bintray.com/google/tensorflow/tensorflow-android](https://bintray.com/google/tensorflow/tensorflow-android).
+You may replace the `+` with an explicit version label if you wish to
+use a specific release of TensorFlow in your app.
+
 To build the libraries yourself (if, for example, you want to support custom
 TensorFlow operators), pick your preferred approach below:
 

tensorflow/contrib/crf/python/ops/crf.py

@@ -41,11 +41,11 @@
 
 import numpy as np
 
-from tensorflow.contrib.rnn.python.ops import core_rnn_cell
 from tensorflow.python.framework import dtypes
 from tensorflow.python.ops import array_ops
 from tensorflow.python.ops import math_ops
 from tensorflow.python.ops import rnn
+from tensorflow.python.ops import rnn_cell
 from tensorflow.python.ops import variable_scope as vs
 
 __all__ = [
@@ -225,7 +225,7 @@ def crf_binary_score(tag_indices, sequence_lengths, transition_params):
   return binary_scores
 
 
-class CrfForwardRnnCell(core_rnn_cell.RNNCell):
+class CrfForwardRnnCell(rnn_cell.RNNCell):
   """Computes the alpha values in a linear-chain CRF.
 
   See http://www.cs.columbia.edu/~mcollins/fb.pdf for reference.

tensorflow/contrib/cudnn_rnn/python/kernel_tests/cudnn_rnn_ops_benchmark.py

@@ -22,7 +22,6 @@
 
 from tensorflow.contrib.cudnn_rnn.python.ops import cudnn_rnn_ops
 from tensorflow.contrib.rnn.python.ops import core_rnn
-from tensorflow.contrib.rnn.python.ops import core_rnn_cell_impl
 from tensorflow.contrib.rnn.python.ops import lstm_ops
 from tensorflow.python.client import session
 from tensorflow.python.framework import dtypes
@@ -31,6 +30,7 @@
 from tensorflow.python.ops import control_flow_ops
 from tensorflow.python.ops import gradients_impl
 from tensorflow.python.ops import init_ops
+from tensorflow.python.ops import rnn_cell
 from tensorflow.python.ops import variables
 from tensorflow.python.platform import test
 
@@ -131,9 +131,9 @@ def benchmarkTfRNNLSTMTraining(self):
         ]
         initializer = init_ops.random_uniform_initializer(-0.01, 0.01, seed=127)
 
-        cell = core_rnn_cell_impl.LSTMCell(
+        cell = rnn_cell.LSTMCell(
             num_units=num_units, initializer=initializer, state_is_tuple=True)
-        multi_cell = core_rnn_cell_impl.MultiRNNCell(
+        multi_cell = rnn_cell.MultiRNNCell(
             [cell() for _ in range(num_layers)])
         outputs, final_state = core_rnn.static_rnn(
             multi_cell, inputs, dtype=dtypes.float32)
@@ -159,7 +159,7 @@ def benchmarkTfRNNLSTMBlockCellTraining(self):
         ]
         cell = lambda: lstm_ops.LSTMBlockCell(num_units=num_units)  # pylint: disable=cell-var-from-loop
 
-        multi_cell = core_rnn_cell_impl.MultiRNNCell(
+        multi_cell = rnn_cell.MultiRNNCell(
             [cell() for _ in range(num_layers)])
         outputs, final_state = core_rnn.static_rnn(
             multi_cell, inputs, dtype=dtypes.float32)

tensorflow/contrib/grid_rnn/python/kernel_tests/grid_rnn_test.py

@@ -21,11 +21,11 @@
 import numpy as np
 
 from tensorflow.contrib.grid_rnn.python.ops import grid_rnn_cell
-from tensorflow.contrib.rnn.python.ops import core_rnn
 from tensorflow.python.framework import dtypes
 from tensorflow.python.ops import array_ops
 from tensorflow.python.ops import init_ops
 from tensorflow.python.ops import nn_ops
+from tensorflow.python.ops import rnn
 from tensorflow.python.ops import variable_scope
 from tensorflow.python.ops import variables
 from tensorflow.python.platform import test
@@ -527,7 +527,7 @@ def testGrid2LSTMCellWithRNN(self):
               dtypes.float32, shape=(batch_size, input_size))
       ]
 
-      outputs, state = core_rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
+      outputs, state = rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
 
     self.assertEqual(len(outputs), len(inputs))
     self.assertEqual(state[0].c.get_shape(), (batch_size, 2))
@@ -569,7 +569,7 @@ def testGrid2LSTMCellReLUWithRNN(self):
           array_ops.placeholder(dtypes.float32, shape=(batch_size, input_size))
       ]
 
-      outputs, state = core_rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
+      outputs, state = rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
 
     self.assertEqual(len(outputs), len(inputs))
     self.assertEqual(state[0].c.get_shape(), (batch_size, 2))
@@ -609,7 +609,7 @@ def testGrid3LSTMCellReLUWithRNN(self):
           array_ops.placeholder(dtypes.float32, shape=(batch_size, input_size))
       ]
 
-      outputs, state = core_rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
+      outputs, state = rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
 
     self.assertEqual(len(outputs), len(inputs))
     self.assertEqual(state[0].c.get_shape(), (batch_size, 2))
@@ -652,7 +652,7 @@ def testGrid1LSTMCellWithRNN(self):
               dtypes.float32, shape=(batch_size, input_size))
       ] + (max_length - 1) * [array_ops.zeros([batch_size, input_size])])
 
-      outputs, state = core_rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
+      outputs, state = rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
 
     self.assertEqual(len(outputs), len(inputs))
     self.assertEqual(state[0].c.get_shape(), (batch_size, 2))
@@ -690,7 +690,7 @@ def testGrid2LSTMCellWithRNNAndDynamicBatchSize(self):
           array_ops.placeholder(dtypes.float32, shape=(None, input_size))
       ]
 
-      outputs, state = core_rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
+      outputs, state = rnn.static_rnn(cell, inputs, dtype=dtypes.float32)
 
     self.assertEqual(len(outputs), len(inputs))
 

tensorflow/contrib/learn/python/learn/estimators/dynamic_rnn_estimator_test.py

@@ -31,7 +31,6 @@
 from tensorflow.contrib.learn.python.learn.estimators import prediction_key
 from tensorflow.contrib.learn.python.learn.estimators import rnn_common
 from tensorflow.contrib.learn.python.learn.estimators import run_config
-from tensorflow.contrib.rnn.python.ops import core_rnn_cell_impl
 from tensorflow.python.client import session
 from tensorflow.python.framework import constant_op
 from tensorflow.python.framework import dtypes
@@ -42,6 +41,7 @@
 from tensorflow.python.ops import lookup_ops
 from tensorflow.python.ops import math_ops
 from tensorflow.python.ops import random_ops
+from tensorflow.python.ops import rnn_cell
 from tensorflow.python.ops import variables
 from tensorflow.python.platform import test
 
@@ -107,7 +107,7 @@ class DynamicRnnEstimatorTest(test.TestCase):
 
   def setUp(self):
     super(DynamicRnnEstimatorTest, self).setUp()
-    self.rnn_cell = core_rnn_cell_impl.BasicRNNCell(self.NUM_RNN_CELL_UNITS)
+    self.rnn_cell = rnn_cell.BasicRNNCell(self.NUM_RNN_CELL_UNITS)
     self.mock_target_column = MockTargetColumn(
         num_label_columns=self.NUM_LABEL_COLUMNS)
 
@@ -312,19 +312,19 @@ def testStateTupleDictConversion(self):
     # A MultiRNNCell of LSTMCells is both a common choice and an interesting
     # test case, because it has two levels of nesting, with an inner class that
     # is not a plain tuple.
-    cell = core_rnn_cell_impl.MultiRNNCell(
-        [core_rnn_cell_impl.LSTMCell(i) for i in cell_sizes])
+    cell = rnn_cell.MultiRNNCell(
+        [rnn_cell.LSTMCell(i) for i in cell_sizes])
     state_dict = {
         dynamic_rnn_estimator._get_state_name(i):
         array_ops.expand_dims(math_ops.range(cell_size), 0)
         for i, cell_size in enumerate([5, 5, 3, 3, 7, 7])
     }
-    expected_state = (core_rnn_cell_impl.LSTMStateTuple(
+    expected_state = (rnn_cell.LSTMStateTuple(
         np.reshape(np.arange(5), [1, -1]), np.reshape(np.arange(5), [1, -1])),
-                      core_rnn_cell_impl.LSTMStateTuple(
+                      rnn_cell.LSTMStateTuple(
                           np.reshape(np.arange(3), [1, -1]),
                           np.reshape(np.arange(3), [1, -1])),
-                      core_rnn_cell_impl.LSTMStateTuple(
+                      rnn_cell.LSTMStateTuple(
                           np.reshape(np.arange(7), [1, -1]),
                           np.reshape(np.arange(7), [1, -1])))
     actual_state = dynamic_rnn_estimator.dict_to_state_tuple(state_dict, cell)

tensorflow/contrib/learn/python/learn/estimators/state_saving_rnn_estimator.py

@@ -26,13 +26,13 @@
 from tensorflow.contrib.learn.python.learn.estimators import estimator
 from tensorflow.contrib.learn.python.learn.estimators import model_fn
 from tensorflow.contrib.learn.python.learn.estimators import rnn_common
-from tensorflow.contrib.rnn.python.ops import core_rnn
 from tensorflow.contrib.training.python.training import sequence_queueing_state_saver as sqss
 from tensorflow.python.framework import dtypes
 from tensorflow.python.framework import ops
 from tensorflow.python.framework import sparse_tensor
 from tensorflow.python.framework import tensor_shape
 from tensorflow.python.ops import array_ops
+from tensorflow.python.ops import rnn
 from tensorflow.python.training import momentum as momentum_opt
 from tensorflow.python.util import nest
 
@@ -64,7 +64,7 @@ def construct_state_saving_rnn(cell,
     final_state: The final state output by the RNN
   """
   with ops.name_scope(scope):
-    rnn_outputs, final_state = core_rnn.static_state_saving_rnn(
+    rnn_outputs, final_state = rnn.static_state_saving_rnn(
         cell=cell,
         inputs=inputs,
         state_saver=state_saver,

tensorflow/contrib/learn/python/learn/ops/seq2seq_ops_test.py

@@ -21,9 +21,9 @@
 import numpy as np
 
 from tensorflow.contrib.learn.python.learn import ops
-from tensorflow.contrib.rnn.python.ops import core_rnn_cell_impl
 from tensorflow.python.framework import dtypes
 from tensorflow.python.ops import array_ops
+from tensorflow.python.ops import rnn_cell
 from tensorflow.python.platform import test
 
 
@@ -82,7 +82,7 @@ def test_rnn_decoder(self):
           array_ops.placeholder(dtypes.float32, [2, 2]) for _ in range(3)
       ]
       encoding = array_ops.placeholder(dtypes.float32, [2, 2])
-      cell = core_rnn_cell_impl.GRUCell(2)
+      cell = rnn_cell.GRUCell(2)
       outputs, states, sampling_outputs, sampling_states = (
           ops.rnn_decoder(decoder_inputs, encoding, cell))
       self.assertEqual(len(outputs), 3)