Remove barrier, add tf.identity where appropriate, and make sure tests pass

inception/inception/slim/ops_test.py

@@ -418,7 +418,7 @@ class DropoutTest(tf.test.TestCase):
     with self.test_session():
       images = tf.random_uniform((5, height, width, 3), seed=1)
       output = ops.dropout(images)
-      self.assertEquals(output.op.name, 'Dropout/dropout/mul_1')
+      self.assertEquals(output.op.name, 'Dropout/dropout/mul')
       output.get_shape().assert_is_compatible_with(images.get_shape())
 
   def testCreateDropoutNoTraining(self):
@@ -599,8 +599,6 @@ class BatchNormTest(tf.test.TestCase):
       output = ops.batch_norm(images, decay=0.1)
       update_ops = tf.get_collection(ops.UPDATE_OPS_COLLECTION)
       with tf.control_dependencies(update_ops):
-        barrier = tf.no_op(name='gradient_barrier')
-        with tf.control_dependencies([barrier]):
         output = tf.identity(output)
       # Initialize all variables
       sess.run(tf.global_variables_initializer())
@@ -630,8 +628,6 @@ class BatchNormTest(tf.test.TestCase):
       output = ops.batch_norm(images, decay=0.1, is_training=False)
       update_ops = tf.get_collection(ops.UPDATE_OPS_COLLECTION)
       with tf.control_dependencies(update_ops):
-        barrier = tf.no_op(name='gradient_barrier')
-        with tf.control_dependencies([barrier]):
         output = tf.identity(output)
       # Initialize all variables
       sess.run(tf.global_variables_initializer())
@@ -665,8 +661,6 @@ class BatchNormTest(tf.test.TestCase):
       output = ops.batch_norm(images, decay=0.1, is_training=False)
       update_ops = tf.get_collection(ops.UPDATE_OPS_COLLECTION)
       with tf.control_dependencies(update_ops):
-        barrier = tf.no_op(name='gradient_barrier')
-        with tf.control_dependencies([barrier]):
         output = tf.identity(output)
       # Initialize all variables
       sess.run(tf.global_variables_initializer())

slim/deployment/model_deploy.py

@@ -379,7 +379,7 @@ def deploy(config,
 
         update_op = tf.group(*update_ops)
         with tf.control_dependencies([update_op]):
-          train_op = total_loss
+          train_op = tf.identity(total_loss, name='train_op')
     else:
       clones_losses = []
       regularization_losses = tf.get_collection(

slim/train_image_classifier.py

@@ -540,7 +540,7 @@ def main(_):
 
     update_op = tf.group(*update_ops)
     with tf.control_dependencies([update_op]):
-      train_tensor = total_loss
+      train_tensor = tf.identity(total_loss, name='train_op')
 
     # Add the summaries from the first clone. These contain the summaries
     # created by model_fn and either optimize_clones() or _gather_clone_loss().