Shift tensorflow/examples over to the new summary ops.

This is a fairly minimal change, some functions have been cleaned up slightly
because the new summary ops dont need manual namespacing.

Behavior is approximately identical.
Change: 136745220

tensorflow/examples/how_tos/reading_data/fully_connected_preloaded.py

@@ -75,7 +75,7 @@ def run_training():
     eval_correct = mnist.evaluation(logits, labels)
 
     # Build the summary operation based on the TF collection of Summaries.
-    summary_op = tf.merge_all_summaries()
+    summary_op = tf.summary.merge_all()
 
     # Create a saver for writing training checkpoints.
     saver = tf.train.Saver()

tensorflow/examples/how_tos/reading_data/fully_connected_preloaded_var.py

@@ -81,7 +81,7 @@ def run_training():
     eval_correct = mnist.evaluation(logits, labels)
 
     # Build the summary operation based on the TF collection of Summaries.
-    summary_op = tf.merge_all_summaries()
+    summary_op = tf.summary.merge_all()
 
     # Create a saver for writing training checkpoints.
     saver = tf.train.Saver()

tensorflow/examples/image_retraining/retrain.py

@@ -647,17 +647,17 @@ def add_input_distortions(flip_left_right, random_crop, random_scale,
   return jpeg_data, distort_result
 
 
-def variable_summaries(var, name):
+def variable_summaries(var):
   """Attach a lot of summaries to a Tensor (for TensorBoard visualization)."""
   with tf.name_scope('summaries'):
     mean = tf.reduce_mean(var)
-    tf.scalar_summary('mean/' + name, mean)
+    tf.summary.scalar('mean', mean)
     with tf.name_scope('stddev'):
       stddev = tf.sqrt(tf.reduce_mean(tf.square(var - mean)))
-    tf.scalar_summary('stddev/' + name, stddev)
-    tf.scalar_summary('max/' + name, tf.reduce_max(var))
-    tf.scalar_summary('min/' + name, tf.reduce_min(var))
-    tf.histogram_summary(name, var)
+    tf.summary.scalar('stddev', stddev)
+    tf.summary.scalar('max', tf.reduce_max(var))
+    tf.summary.scalar('min', tf.reduce_min(var))
+    tf.summary.histogram('histogram', var)
 
 
 def add_final_training_ops(class_count, final_tensor_name, bottleneck_tensor):
@@ -695,23 +695,23 @@ def add_final_training_ops(class_count, final_tensor_name, bottleneck_tensor):
   with tf.name_scope(layer_name):
     with tf.name_scope('weights'):
       layer_weights = tf.Variable(tf.truncated_normal([BOTTLENECK_TENSOR_SIZE, class_count], stddev=0.001), name='final_weights')
-      variable_summaries(layer_weights, layer_name + '/weights')
+      variable_summaries(layer_weights)
     with tf.name_scope('biases'):
       layer_biases = tf.Variable(tf.zeros([class_count]), name='final_biases')
-      variable_summaries(layer_biases, layer_name + '/biases')
+      variable_summaries(layer_biases)
     with tf.name_scope('Wx_plus_b'):
       logits = tf.matmul(bottleneck_input, layer_weights) + layer_biases
-      tf.histogram_summary(layer_name + '/pre_activations', logits)
+      tf.summary.histogram('pre_activations', logits)
 
   final_tensor = tf.nn.softmax(logits, name=final_tensor_name)
-  tf.histogram_summary(final_tensor_name + '/activations', final_tensor)
+  tf.summary.histogram('activations', final_tensor)
 
   with tf.name_scope('cross_entropy'):
     cross_entropy = tf.nn.softmax_cross_entropy_with_logits(
       logits, ground_truth_input)
     with tf.name_scope('total'):
       cross_entropy_mean = tf.reduce_mean(cross_entropy)
-    tf.scalar_summary('cross entropy', cross_entropy_mean)
+  tf.summary.scalar('cross_entropy', cross_entropy_mean)
 
   with tf.name_scope('train'):
     train_step = tf.train.GradientDescentOptimizer(FLAGS.learning_rate).minimize(
@@ -738,7 +738,7 @@ def add_evaluation_step(result_tensor, ground_truth_tensor):
         tf.argmax(ground_truth_tensor, 1))
     with tf.name_scope('accuracy'):
       evaluation_step = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
-    tf.scalar_summary('accuracy', evaluation_step)
+  tf.summary.scalar('accuracy', evaluation_step)
   return evaluation_step
 
 
@@ -792,7 +792,7 @@ def main(_):
   evaluation_step = add_evaluation_step(final_tensor, ground_truth_input)
 
   # Merge all the summaries and write them out to /tmp/retrain_logs (by default)
-  merged = tf.merge_all_summaries()
+  merged = tf.summary.merge_all()
   train_writer = tf.train.SummaryWriter(FLAGS.summaries_dir + '/train',
                                         sess.graph)
   validation_writer = tf.train.SummaryWriter(FLAGS.summaries_dir + '/validation')

tensorflow/examples/tutorials/mnist/fully_connected_feed.py

@@ -139,7 +139,7 @@ def run_training():
     eval_correct = mnist.evaluation(logits, labels_placeholder)
 
     # Build the summary Tensor based on the TF collection of Summaries.
-    summary = tf.merge_all_summaries()
+    summary = tf.summary.merge_all()
 
     # Add the variable initializer Op.
     init = tf.initialize_all_variables()

tensorflow/examples/tutorials/mnist/mnist.py

@@ -118,7 +118,7 @@ def training(loss, learning_rate):
     train_op: The Op for training.
   """
   # Add a scalar summary for the snapshot loss.
-  tf.scalar_summary(loss.op.name, loss)
+  tf.summary.scalar('loss', loss)
   # Create the gradient descent optimizer with the given learning rate.
   optimizer = tf.train.GradientDescentOptimizer(learning_rate)
   # Create a variable to track the global step.

tensorflow/examples/tutorials/mnist/mnist_with_summaries.py

@@ -40,7 +40,6 @@ def train():
                                     fake_data=FLAGS.fake_data)
 
   sess = tf.InteractiveSession()
-
   # Create a multilayer model.
 
   # Input placeholders
@@ -50,7 +49,7 @@ def train():
 
   with tf.name_scope('input_reshape'):
     image_shaped_input = tf.reshape(x, [-1, 28, 28, 1])
-    tf.image_summary('input', image_shaped_input, 10)
+    tf.summary.image('input', image_shaped_input, 10)
 
   # We can't initialize these variables to 0 - the network will get stuck.
   def weight_variable(shape):
@@ -63,17 +62,17 @@ def train():
     initial = tf.constant(0.1, shape=shape)
     return tf.Variable(initial)
 
-  def variable_summaries(var, name):
-    """Attach a lot of summaries to a Tensor."""
+  def variable_summaries(var):
+    """Attach a lot of summaries to a Tensor (for TensorBoard visualization)."""
     with tf.name_scope('summaries'):
       mean = tf.reduce_mean(var)
-      tf.scalar_summary('mean/' + name, mean)
+      tf.summary.scalar('mean', mean)
       with tf.name_scope('stddev'):
         stddev = tf.sqrt(tf.reduce_mean(tf.square(var - mean)))
-      tf.scalar_summary('stddev/' + name, stddev)
-      tf.scalar_summary('max/' + name, tf.reduce_max(var))
-      tf.scalar_summary('min/' + name, tf.reduce_min(var))
-      tf.histogram_summary(name, var)
+      tf.summary.scalar('stddev', stddev)
+      tf.summary.scalar('max', tf.reduce_max(var))
+      tf.summary.scalar('min', tf.reduce_min(var))
+      tf.summary.histogram('histogram', var)
 
   def nn_layer(input_tensor, input_dim, output_dim, layer_name, act=tf.nn.relu):
     """Reusable code for making a simple neural net layer.
@@ -87,22 +86,22 @@ def train():
       # This Variable will hold the state of the weights for the layer
       with tf.name_scope('weights'):
         weights = weight_variable([input_dim, output_dim])
-        variable_summaries(weights, layer_name + '/weights')
+        variable_summaries(weights)
       with tf.name_scope('biases'):
         biases = bias_variable([output_dim])
-        variable_summaries(biases, layer_name + '/biases')
+        variable_summaries(biases)
       with tf.name_scope('Wx_plus_b'):
         preactivate = tf.matmul(input_tensor, weights) + biases
-        tf.histogram_summary(layer_name + '/pre_activations', preactivate)
+        tf.summary.histogram('pre_activations', preactivate)
       activations = act(preactivate, name='activation')
-      tf.histogram_summary(layer_name + '/activations', activations)
+      tf.summary.histogram('activations', activations)
       return activations
 
   hidden1 = nn_layer(x, 784, 500, 'layer1')
 
   with tf.name_scope('dropout'):
     keep_prob = tf.placeholder(tf.float32)
-    tf.scalar_summary('dropout_keep_probability', keep_prob)
+    tf.summary.scalar('dropout_keep_probability', keep_prob)
     dropped = tf.nn.dropout(hidden1, keep_prob)
 
   # Do not apply softmax activation yet, see below.
@@ -122,7 +121,7 @@ def train():
     diff = tf.nn.softmax_cross_entropy_with_logits(y, y_)
     with tf.name_scope('total'):
       cross_entropy = tf.reduce_mean(diff)
-    tf.scalar_summary('cross entropy', cross_entropy)
+  tf.summary.scalar('cross_entropy', cross_entropy)
 
   with tf.name_scope('train'):
     train_step = tf.train.AdamOptimizer(FLAGS.learning_rate).minimize(
@@ -133,10 +132,10 @@ def train():
       correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
     with tf.name_scope('accuracy'):
       accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
-    tf.scalar_summary('accuracy', accuracy)
+  tf.summary.scalar('accuracy', accuracy)
 
   # Merge all the summaries and write them out to /tmp/mnist_logs (by default)
-  merged = tf.merge_all_summaries()
+  merged = tf.summary.merge_all()
   train_writer = tf.train.SummaryWriter(FLAGS.summaries_dir + '/train',
                                         sess.graph)
   test_writer = tf.train.SummaryWriter(FLAGS.summaries_dir + '/test')

tensorflow/g3doc/how_tos/distributed/index.md

@@ -226,7 +226,7 @@ def main(_):
           loss, global_step=global_step)
 
       saver = tf.train.Saver()
-      summary_op = tf.merge_all_summaries()
+      summary_op = tf.summary.merge_all()
       init_op = tf.initialize_all_variables()
 
     # Create a "supervisor", which oversees the training process.

tensorflow/g3doc/how_tos/meta_graph/index.md

@@ -219,7 +219,7 @@ Here are some of the typical usage models:
                                                             name="xentropy")
     loss = tf.reduce_mean(cross_entropy, name="xentropy_mean")
 
-    tf.scalar_summary(loss.op.name, loss)
+    tf.summary.scalar('loss', loss)
     # Creates the gradient descent optimizer with the given learning rate.
     optimizer = tf.train.GradientDescentOptimizer(0.01)
 

tensorflow/g3doc/how_tos/summaries_and_tensorboard/index.md

@@ -24,12 +24,12 @@ lifecycle for summary data within TensorBoard.
 
 First, create the TensorFlow graph that you'd like to collect summary
 data from, and decide which nodes you would like to annotate with
-[summary operations](../../api_docs/python/train.md#summary-operations).
+[summary operations](../../api_docs/python/summary.md).
 
 For example, suppose you are training a convolutional neural network for
 recognizing MNIST digits. You'd like to record how the learning rate
 varies over time, and how the objective function is changing. Collect these by
-attaching [`scalar_summary`](../../api_docs/python/train.md#scalar_summary) ops
+attaching [`scalar_summary`](../../api_docs/python/summary.md#scalar) ops
 to the nodes that output the learning rate and loss respectively. Then, give
 each `scalar_summary` a meaningful `tag`, like `'learning rate'` or `'loss
 function'`.
@@ -37,18 +37,18 @@ function'`.
 Perhaps you'd also like to visualize the distributions of activations coming
 off a particular layer, or the distribution of gradients or weights. Collect
 this data by attaching
-[`histogram_summary`](../../api_docs/python/train.md#histogram_summary) ops to
+[`histogram_summary`](../../api_docs/python/summary.md#histogram) ops to
 the gradient outputs and to the variable that holds your weights, respectively.
 
 For details on all of the summary operations available, check out the docs on
-[summary operations](../../api_docs/python/train.md#summary-operations).
+[summary operations](../../api_docs/python/summary.md).
 
 Operations in TensorFlow don't do anything until you run them, or an op that
 depends on their output. And the summary nodes that we've just created are
 peripheral to your graph: none of the ops you are currently running depend on
 them. So, to generate summaries, we need to run all of these summary nodes.
 Managing them by hand would be tedious, so use
-[`tf.merge_all_summaries`](../../api_docs/python/train.md#merge_all_summaries)
+[`tf.summary.merge_all`](../../api_docs/python/summary.md#merge_all)
 to combine them into a single op that generates all the summary data.
 
 Then, you can just run the merged summary op, which will generate a serialized
@@ -79,17 +79,17 @@ training. The code below is an excerpt; full source is
 [here](https://www.tensorflow.org/code/tensorflow/examples/tutorials/mnist/mnist_with_summaries.py).
 
 ```python
-def variable_summaries(var, name):
-  """Attach a lot of summaries to a Tensor."""
+def variable_summaries(var):
+  """Attach a lot of summaries to a Tensor (for TensorBoard visualization)."""
   with tf.name_scope('summaries'):
     mean = tf.reduce_mean(var)
-    tf.scalar_summary('mean/' + name, mean)
+    tf.summary.scalar('mean', mean)
     with tf.name_scope('stddev'):
       stddev = tf.sqrt(tf.reduce_mean(tf.square(var - mean)))
-    tf.scalar_summary('stddev/' + name, stddev)
-    tf.scalar_summary('max/' + name, tf.reduce_max(var))
-    tf.scalar_summary('min/' + name, tf.reduce_min(var))
-    tf.histogram_summary(name, var)
+    tf.summary.scalar('stddev', stddev)
+    tf.summary.scalar('max', tf.reduce_max(var))
+    tf.summary.scalar('min', tf.reduce_min(var))
+    tf.summary.histogram('histogram', var)
 
 def nn_layer(input_tensor, input_dim, output_dim, layer_name, act=tf.nn.relu):
   """Reusable code for making a simple neural net layer.
@@ -103,31 +103,42 @@ def nn_layer(input_tensor, input_dim, output_dim, layer_name, act=tf.nn.relu):
     # This Variable will hold the state of the weights for the layer
     with tf.name_scope('weights'):
       weights = weight_variable([input_dim, output_dim])
-      variable_summaries(weights, layer_name + '/weights')
+      variable_summaries(weights)
     with tf.name_scope('biases'):
       biases = bias_variable([output_dim])
-      variable_summaries(biases, layer_name + '/biases')
+      variable_summaries(biases)
     with tf.name_scope('Wx_plus_b'):
       preactivate = tf.matmul(input_tensor, weights) + biases
-      tf.histogram_summary(layer_name + '/pre_activations', preactivate)
-    activations = act(preactivate, 'activation')
-    tf.histogram_summary(layer_name + '/activations', activations)
+      tf.summary.histogram('pre_activations', preactivate)
+    activations = act(preactivate, name='activation')
+    tf.summary.histogram('activations', activations)
     return activations
 
 hidden1 = nn_layer(x, 784, 500, 'layer1')
 
 with tf.name_scope('dropout'):
   keep_prob = tf.placeholder(tf.float32)
-  tf.scalar_summary('dropout_keep_probability', keep_prob)
+  tf.summary.scalar('dropout_keep_probability', keep_prob)
   dropped = tf.nn.dropout(hidden1, keep_prob)
 
-y = nn_layer(dropped, 500, 10, 'layer2', act=tf.nn.softmax)
+# Do not apply softmax activation yet, see below.
+y = nn_layer(dropped, 500, 10, 'layer2', act=tf.identity)
 
 with tf.name_scope('cross_entropy'):
-  diff = y_ * tf.log(y)
+  # The raw formulation of cross-entropy,
+  #
+  # tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(tf.softmax(y)),
+  #                               reduction_indices=[1]))
+  #
+  # can be numerically unstable.
+  #
+  # So here we use tf.nn.softmax_cross_entropy_with_logits on the
+  # raw outputs of the nn_layer above, and then average across
+  # the batch.
+  diff = tf.nn.softmax_cross_entropy_with_logits(y, y_)
   with tf.name_scope('total'):
-    cross_entropy = -tf.reduce_mean(diff)
-  tf.scalar_summary('cross entropy', cross_entropy)
+    cross_entropy = tf.reduce_mean(diff)
+tf.summary.scalar('cross_entropy', cross_entropy)
 
 with tf.name_scope('train'):
   train_step = tf.train.AdamOptimizer(FLAGS.learning_rate).minimize(
@@ -138,10 +149,10 @@ with tf.name_scope('accuracy'):
     correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
   with tf.name_scope('accuracy'):
     accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
-  tf.scalar_summary('accuracy', accuracy)
+tf.summary.scalar('accuracy', accuracy)
 
 # Merge all the summaries and write them out to /tmp/mnist_logs (by default)
-merged = tf.merge_all_summaries()
+merged = tf.summary.merge_all()
 train_writer = tf.train.SummaryWriter(FLAGS.summaries_dir + '/train',
                                       sess.graph)
 test_writer = tf.train.SummaryWriter(FLAGS.summaries_dir + '/test')

tensorflow/g3doc/how_tos/supervisor/index.md

@@ -130,7 +130,7 @@ For example this code runs the summary op every 100 steps in the training loop:
 ```python
   ...create graph...
   my_train_op = ...
-  my_summary_op = tf.merge_all_summaries()
+  my_summary_op = tf.summary.merge_all()
 
   sv = tf.Supervisor(logdir="/my/training/directory",
                      summary_op=None) # Do not run the summary service
@@ -317,7 +317,7 @@ constructor:
    `tf.GraphKeys.SUMMARY_OP` [graph
    collection](../../api_docs/python/framework#Graph.add_to_collection).  If
    the collection is empty the supervisor creates an op that aggregates all
-   summaries in the graph using `tf.merge_all_summaries()`.
+   summaries in the graph using `tf.summary.merge_all()`.
 
    Passing `None` disables the summary service.
 
@@ -404,5 +404,3 @@ Checkpoint recovery is controlled by the following keyword arguments to the
    ready op the first time, to initialize local variables and tables.
 
  * `saver`: (see above).  Saver object used to load the checkpoint.
-
-