Release 1.5.1 for TF1.2

Author: zsdonghao

README.md

@@ -18,6 +18,7 @@
 TensorLayer is a deep learning and reinforcement learning library based on [Google TensorFlow](https://www.tensorflow.org). It provides rich data pre-processing,  training, post-processing and serving modules that help researchers and engineers in building complex machine learning workflows.  
 
 # What's New
+* Support TensorFlow 1.2
 * Join [Slack](https://join.slack.com/tensorlayer/shared_invite/MTk1MjM0NDk5MDg5LTE0OTcwOTQyNTEtODFjY2QzYjdmZQ) Now.
 * [Attention Seq2seq](https://github.com/zsdonghao/tensorlayer/issues/164) help wanted.
 * Support [Sub-pixel Convolution](http://tensorlayer.readthedocs.io/en/latest/modules/layers.html#super-resolution-layer) for Super-resolution.

docs/conf.py

@@ -67,9 +67,9 @@
 # built documents.
 #
 # The short X.Y version.
-version = '1.5.0'
+version = '1.5.1'
 # The full version, including alpha/beta/rc tags.
-release = '1.5.0'
+release = '1.5.1'
 
 # The language for content autogenerated by Sphinx. Refer to documentation
 # for a list of supported languages.
@@ -143,7 +143,7 @@
 # The name for this set of Sphinx documents.
 # "<project> v<release> documentation" by default.
 #
-# html_title = 'TensorLayer v1.5.0'
+# html_title = 'TensorLayer v1.5.1'
 
 # A shorter title for the navigation bar.  Default is the same as html_title.
 #

example/_tf0.12/tutorial_dynamic_rnn.py

@@ -26,7 +26,8 @@
 X[1,6:] = 0
 X_lengths = [10, 6]
 
-cell = tf.nn.rnn_cell.LSTMCell(num_units=64, state_is_tuple=True)
+cell =tf.contrib.rnn.BasicLSTMCell(num_units=64, state_is_tuple=True)
+# cell = tf.nn.rnn_cell.LSTMCell(num_units=64, state_is_tuple=True) # TF 0.12
 
 outputs, last_states = tf.nn.dynamic_rnn(
     cell=cell,
@@ -47,7 +48,8 @@
 ## 2. How to control the initial state
 batch_size = X.shape[0]
 with tf.variable_scope('name') as vs:
-    cell = tf.nn.rnn_cell.LSTMCell(num_units=64, state_is_tuple=True)
+    cell =tf.contrib.rnn.BasicLSTMCell(num_units=64, state_is_tuple=True)
+    # cell = tf.nn.rnn_cell.LSTMCell(num_units=64, state_is_tuple=True) # TF 0.12
     initial_state = cell.zero_state(batch_size, dtype=tf.float64)#"float")
     outputs, last_states = tf.nn.dynamic_rnn(
                         cell=cell,
@@ -84,11 +86,12 @@ def retrieve_seq_length_op(data):
     return length
 
 sequence_length = retrieve_seq_length_op(
-            incoming if isinstance(X, tf.Tensor) else tf.pack(X))
+            incoming if isinstance(X, tf.Tensor) else tf.stack(X))#tf.pack(X))
 
 batch_size = X.shape[0]
 with tf.variable_scope('name2') as vs: #, initializer=tf.constant_initializer(value=0.1)) as vs:
-    cell = tf.nn.rnn_cell.LSTMCell(num_units=64, state_is_tuple=True)
+    cell =tf.contrib.rnn.BasicLSTMCell(num_units=64, state_is_tuple=True)
+    # cell = tf.nn.rnn_cell.LSTMCell(num_units=64, state_is_tuple=True) # TF 0.12
     initial_state = cell.zero_state(batch_size, dtype=tf.float64)#"float")
     outputs, last_states = tf.nn.dynamic_rnn(
                         cell=cell,
@@ -106,11 +109,11 @@ def retrieve_seq_length_op(data):
 # print(sequence_length)
 # exit()
 # automatically get the last output
-outputs = tf.transpose(tf.pack(outputs), [1, 0, 2])
+outputs = tf.transpose(tf.stack(outputs), [1, 0, 2]) #outputs = tf.transpose(tf.pack(outputs), [1, 0, 2])
 last_outputs = advanced_indexing_op(outputs, sequence_length)
 last_states = result[0]["last_states"]
 sess = tf.Session()
-sess.run(tf.initialize_all_variables())
+tl.layers.initialize_global_variables(sess)
 # print('last outputs',sess.run(last_outputs)) # (2, 64)  # TO DO
 # print('last lstm states',last_states, last_states.c.shape, last_states.h.shape)