tfdbg: example for debugging tf-learn Estimator.fit() & minor changes

$ blaze build -c opt third_party/tensorflow/python/debug:debug_tflearn_iris && blaze-bin/third_party/tensorflow/python/debug/debug_tflearn_iris --debug --train_steps=1

Minor changes made in this CL:
* Fix a bug in local_cli_wrapper related to computing fetch and feed summary strings.
* Minor tweaks in debug/examples/README.md
* Curses CLI: Add scroll direction information for UI clarity
Change: 138910196

8 files changed, 309 insertions, 32 deletions

diff --git a/tensorflow/python/debug/BUILD b/tensorflow/python/debug/BUILD
index 0410b30db2..9d44840c88 100644
--- a/tensorflow/python/debug/BUILD
+++ b/tensorflow/python/debug/BUILD
@@ -22,6 +22,7 @@ py_library(
     deps = [
         ":debug_data",
         ":debug_utils",
+        ":hooks",
         ":local_cli_wrapper",
     ],
 )
@@ -116,6 +117,16 @@ py_library(
     ],
 )
 
+py_library(
+    name = "hooks",
+    srcs = ["wrappers/hooks.py"],
+    srcs_version = "PY2AND3",
+    deps = [
+        ":local_cli_wrapper",
+        "//tensorflow/python:session",
+    ],
+)
+
 py_binary(
     name = "debug_fibonacci",
     srcs = ["examples/debug_fibonacci.py"],
@@ -140,6 +151,13 @@ py_binary(
     ],
 )
 
+py_binary(
+    name = "debug_tflearn_iris",
+    srcs = ["examples/debug_tflearn_iris.py"],
+    srcs_version = "PY2AND3",
+    deps = ["//tensorflow:tensorflow_py"],
+)
+
 py_test(
     name = "debug_data_test",
     size = "small",
diff --git a/tensorflow/python/debug/__init__.py b/tensorflow/python/debug/__init__.py
index fd7863e6d3..812c31ed16 100644
--- a/tensorflow/python/debug/__init__.py
+++ b/tensorflow/python/debug/__init__.py
@@ -28,4 +28,5 @@ from tensorflow.python.debug.debug_utils import add_debug_tensor_watch
 from tensorflow.python.debug.debug_utils import watch_graph
 from tensorflow.python.debug.debug_utils import watch_graph_with_blacklists
 
+from tensorflow.python.debug.wrappers.hooks import LocalCLIDebugHook
 from tensorflow.python.debug.wrappers.local_cli_wrapper import LocalCLIDebugWrapperSession
diff --git a/tensorflow/python/debug/cli/curses_ui.py b/tensorflow/python/debug/cli/curses_ui.py
index 8e3d069aa0..4dacc6c3c5 100644
--- a/tensorflow/python/debug/cli/curses_ui.py
+++ b/tensorflow/python/debug/cli/curses_ui.py
@@ -937,10 +937,19 @@ class CursesUI(object):
 
     if self._output_pad_height > self._output_pad_screen_height + 1:
       # Display information about the scrolling of tall screen output.
-      self._scroll_info = "--- Scroll: %.2f%% " % (100.0 * (min(
+      scroll_percentage = 100.0 * (min(
           1.0,
           float(self._output_pad_row) /
-          (self._output_pad_height - self._output_pad_screen_height - 1))))
+          (self._output_pad_height - self._output_pad_screen_height - 1)))
+      if self._output_pad_row == 0:
+        scroll_directions = " (PgDn)"
+      elif self._output_pad_row >= (
+          self._output_pad_height - self._output_pad_screen_height - 1):
+        scroll_directions = " (PgUp)"
+      else:
+        scroll_directions = " (PgDn/PgUp)"
+      self._scroll_info = "--- Scroll%s: %.2f%% " % (scroll_directions,
+                                                     scroll_percentage)
 
       self._output_array_pointer_indices = self._show_array_indices()
 
diff --git a/tensorflow/python/debug/cli/curses_ui_test.py b/tensorflow/python/debug/cli/curses_ui_test.py
index 94a80cfac6..8c4b150827 100644
--- a/tensorflow/python/debug/cli/curses_ui_test.py
+++ b/tensorflow/python/debug/cli/curses_ui_test.py
@@ -320,19 +320,19 @@ class CursesTest(test_util.TensorFlowTestCase):
     self.assertEqual(["bar"] * 60, ui.wrapped_outputs[0].lines[:60])
 
     # Initial scroll: At the top.
-    self.assertIn("Scroll: 0.00%", ui.scroll_messages[0])
+    self.assertIn("Scroll (PgDn): 0.00%", ui.scroll_messages[0])
 
     # After 1st scrolling (PageDown).
     # The screen output shouldn't have changed. Only the viewport should.
     self.assertEqual(["bar"] * 60, ui.unwrapped_outputs[0].lines)
     self.assertEqual(["bar"] * 60, ui.wrapped_outputs[0].lines[:60])
-    self.assertIn("Scroll: 1.69%", ui.scroll_messages[1])
+    self.assertIn("Scroll (PgDn/PgUp): 1.69%", ui.scroll_messages[1])
 
     # After 2nd scrolling (PageDown).
-    self.assertIn("Scroll: 3.39%", ui.scroll_messages[2])
+    self.assertIn("Scroll (PgDn/PgUp): 3.39%", ui.scroll_messages[2])
 
     # After 3rd scrolling (PageUp).
-    self.assertIn("Scroll: 1.69%", ui.scroll_messages[3])
+    self.assertIn("Scroll (PgDn/PgUp): 1.69%", ui.scroll_messages[3])
 
   def testCutOffTooManyOutputLines(self):
     ui = MockCursesUI(
@@ -375,16 +375,16 @@ class CursesTest(test_util.TensorFlowTestCase):
     self.assertEqual(["bar"] * 60, ui.wrapped_outputs[0].lines[:60])
 
     # Initial scroll: At the top.
-    self.assertIn("Scroll: 0.00%", ui.scroll_messages[0])
+    self.assertIn("Scroll (PgDn): 0.00%", ui.scroll_messages[0])
 
     # After 1st scrolling (End).
-    self.assertIn("Scroll: 100.00%", ui.scroll_messages[1])
+    self.assertIn("Scroll (PgUp): 100.00%", ui.scroll_messages[1])
 
     # After 2nd scrolling (End).
-    self.assertIn("Scroll: 100.00%", ui.scroll_messages[2])
+    self.assertIn("Scroll (PgUp): 100.00%", ui.scroll_messages[2])
 
     # After 3rd scrolling (Hhome).
-    self.assertIn("Scroll: 0.00%", ui.scroll_messages[3])
+    self.assertIn("Scroll (PgDn): 0.00%", ui.scroll_messages[3])
 
   def testRunUIWithInitCmd(self):
     """Run UI with an initial command specified."""
@@ -398,7 +398,7 @@ class CursesTest(test_util.TensorFlowTestCase):
 
     self.assertEqual(["bar"] * 60, ui.unwrapped_outputs[0].lines)
     self.assertEqual(["bar"] * 60, ui.wrapped_outputs[0].lines[:60])
-    self.assertIn("Scroll: 0.00%", ui.scroll_messages[0])
+    self.assertIn("Scroll (PgDn): 0.00%", ui.scroll_messages[0])
 
   def testCompileHelpWithoutHelpIntro(self):
     ui = MockCursesUI(
@@ -571,7 +571,7 @@ class CursesTest(test_util.TensorFlowTestCase):
 
     # The 1st scroll info should contain scrolling, because the screen size
     # is less than the number of lines in the output.
-    self.assertIn("Scroll: 0.00%", ui.scroll_messages[0])
+    self.assertIn("Scroll (PgDn): 0.00%", ui.scroll_messages[0])
 
   def testTabCompletionWithCommonPrefix(self):
     # Type "b" and trigger tab completion.
@@ -973,77 +973,80 @@ class CursesTest(test_util.TensorFlowTestCase):
         0: None,
         -1: [1, 0]
     }, ui.output_array_pointer_indices[0])
-    self.assertIn(" Scroll: 0.00% -[1,0] ", ui.scroll_messages[0])
+    self.assertIn(" Scroll (PgDn): 0.00% -[1,0] ", ui.scroll_messages[0])
 
     # Scrolled down one line.
     self.assertEqual({
         0: None,
         -1: [2, 0]
     }, ui.output_array_pointer_indices[1])
-    self.assertIn(" Scroll: 16.67% -[2,0] ", ui.scroll_messages[1])
+    self.assertIn(" Scroll (PgDn/PgUp): 16.67% -[2,0] ", ui.scroll_messages[1])
 
     # Scrolled down one line.
     self.assertEqual({
         0: [0, 0],
         -1: [3, 0]
     }, ui.output_array_pointer_indices[2])
-    self.assertIn(" Scroll: 33.33% [0,0]-[3,0] ", ui.scroll_messages[2])
+    self.assertIn(" Scroll (PgDn/PgUp): 33.33% [0,0]-[3,0] ",
+                  ui.scroll_messages[2])
 
     # Scrolled down one line.
     self.assertEqual({
         0: [1, 0],
         -1: [4, 0]
     }, ui.output_array_pointer_indices[3])
-    self.assertIn(" Scroll: 50.00% [1,0]-[4,0] ", ui.scroll_messages[3])
+    self.assertIn(" Scroll (PgDn/PgUp): 50.00% [1,0]-[4,0] ",
+                  ui.scroll_messages[3])
 
     # Scroll to the bottom.
     self.assertEqual({
         0: [4, 0],
         -1: None
     }, ui.output_array_pointer_indices[4])
-    self.assertIn(" Scroll: 100.00% [4,0]- ", ui.scroll_messages[4])
+    self.assertIn(" Scroll (PgUp): 100.00% [4,0]- ", ui.scroll_messages[4])
 
     # Attempt to scroll beyond the bottom should lead to no change.
     self.assertEqual({
         0: [4, 0],
         -1: None
     }, ui.output_array_pointer_indices[5])
-    self.assertIn(" Scroll: 100.00% [4,0]- ", ui.scroll_messages[5])
+    self.assertIn(" Scroll (PgUp): 100.00% [4,0]- ", ui.scroll_messages[5])
 
     # Scrolled up one line.
     self.assertEqual({
         0: [3, 0],
         -1: None
     }, ui.output_array_pointer_indices[6])
-    self.assertIn(" Scroll: 83.33% [3,0]- ", ui.scroll_messages[6])
+    self.assertIn(" Scroll (PgDn/PgUp): 83.33% [3,0]- ", ui.scroll_messages[6])
 
     # Scrolled up one line.
     self.assertEqual({
         0: [2, 0],
         -1: None
     }, ui.output_array_pointer_indices[7])
-    self.assertIn(" Scroll: 66.67% [2,0]- ", ui.scroll_messages[7])
+    self.assertIn(" Scroll (PgDn/PgUp): 66.67% [2,0]- ", ui.scroll_messages[7])
 
     # Scrolled up one line.
     self.assertEqual({
         0: [1, 0],
         -1: [4, 0]
     }, ui.output_array_pointer_indices[8])
-    self.assertIn(" Scroll: 50.00% [1,0]-[4,0] ", ui.scroll_messages[8])
+    self.assertIn(" Scroll (PgDn/PgUp): 50.00% [1,0]-[4,0] ",
+                  ui.scroll_messages[8])
 
     # Scroll to the top.
     self.assertEqual({
         0: None,
         -1: [1, 0]
     }, ui.output_array_pointer_indices[9])
-    self.assertIn(" Scroll: 0.00% -[1,0] ", ui.scroll_messages[9])
+    self.assertIn(" Scroll (PgDn): 0.00% -[1,0] ", ui.scroll_messages[9])
 
     # Attempt to scroll pass the top limit should lead to no change.
     self.assertEqual({
         0: None,
         -1: [1, 0]
     }, ui.output_array_pointer_indices[10])
-    self.assertIn(" Scroll: 0.00% -[1,0] ", ui.scroll_messages[10])
+    self.assertIn(" Scroll (PgDn): 0.00% -[1,0] ", ui.scroll_messages[10])
 
   def testScrollTensorByValidIndices(self):
     """Test scrolling to specified (valid) indices in a tensor."""
diff --git a/tensorflow/python/debug/examples/README.md b/tensorflow/python/debug/examples/README.md
index 8719dea3f7..b58c7516e9 100644
--- a/tensorflow/python/debug/examples/README.md
+++ b/tensorflow/python/debug/examples/README.md
@@ -2,11 +2,13 @@
 
 **(Under development, subject to change)**
 
-This tutorial showcases how to use the TensorFlow Debugger (**tfdbg**) to debug
-a frequently encountered problem in TensorFlow model development: bad numerical
-values (`nan`s and `inf`s) causing training to fail.
+This tutorial showcases the features of TensorFlow Debugger (**tfdbg**)
+command-line interface.
+It contains an example of how to debug a frequently encountered problem in
+TensorFlow model development: bad numerical values (`nan`s and `inf`s) causing
+training to fail.
 
-To observe the issue, run the following code without the debugger:
+To **observe** such an issue, run the following code without the debugger:
 
 ```none
 bazel build -c opt tensorflow/python/debug:debug_mnist && \
@@ -345,11 +347,43 @@ stuck. Success!
        [tfprof](https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/tfprof)
        and other profiling tools for TensorFlow.
 
-**Q**: _How do I link tfdbg against my Session in Bazel?_
+**Q**: _How do I link tfdbg against my `Session` in Bazel?_
 
 **A**: In your BUILD rule, declare the dependency: `"//tensorflow:tensorflow_py"`.
        In your Python file, add:
-         `from tensorflow.python import debug as tf_debug`
+
+```python
+from tensorflow.python import debug as tf_debug
+
+# Then wrap your TensorFlow Session with the local-CLI wrapper.
+sess = tf_debug.LocalCLIDebugWrapperSession(sess)
+```
+
+**Q**: _Can I use `tfdbg` if I am using tf-learn Estimators, instead of
+managing my own `Session` objects?_
+
+**A**: Currently, `tfdbg` can only debug the `fit()` method of tf-learn
+Estimators. Support for debugging `evaluate()` will come soon. To debug
+`Estimator.fit()`, create a monitor and supply it as an argument. For example:
+
+```python
+from tensorflow.python import debug as tf_debug
+
+# Create a local CLI debug hook and use it as a monitor when calling fit().
+classifier.fit(x=training_set.data,
+               y=training_set.target,
+               steps=1000,
+               monitors=[tf_debug.LocalCLIDebugHook()])
+```
+
+For a detailed [example](https://www.tensorflow.org/code/tensorflow/python/debug/examples/debug_tflearn_iris.py) based on
+[tf-learn's iris tutorial](../../../g3doc/tutorials/tflearn/index.md),
+run:
+
+```none
+bazel build -c opt tensorflow/python/debug:debug_tflearn_iris && \
+    bazel-bin/tensorflow/python/debug/debug_tflearn_iris
+```
 
 **Q**: _Does tfdbg help debugging runtime errors such as shape mismatches?_
 
diff --git a/tensorflow/python/debug/examples/debug_tflearn_iris.py b/tensorflow/python/debug/examples/debug_tflearn_iris.py
new file mode 100644
index 0000000000..4c104b3e0a
--- /dev/null
+++ b/tensorflow/python/debug/examples/debug_tflearn_iris.py
@@ -0,0 +1,115 @@
+# Copyright 2016 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Debug the tf-learn iris example, based on the tf-learn tutorial."""
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import os
+import tempfile
+
+import numpy as np
+from six.moves import urllib
+import tensorflow as tf
+
+from tensorflow.python import debug as tf_debug
+
+flags = tf.app.flags
+FLAGS = flags.FLAGS
+flags.DEFINE_string("data_dir", "/tmp/iris_data",
+                    "Directory to save the training and test data in.")
+flags.DEFINE_string("model_dir", "", "Directory to save the trained model in.")
+flags.DEFINE_integer("train_steps", 10, "Number of steps to run trainer.")
+flags.DEFINE_boolean("debug", False,
+                     "Use debugger to track down bad values during training")
+
+# URLs to download data sets from, if necessary.
+IRIS_TRAINING_DATA_URL = "https://raw.githubusercontent.com/tensorflow/tensorflow/master/tensorflow/examples/tutorials/monitors/iris_training.csv"
+IRIS_TEST_DATA_URL = "https://raw.githubusercontent.com/tensorflow/tensorflow/master/tensorflow/examples/tutorials/monitors/iris_test.csv"
+
+
+def maybe_download_data():
+  """Download data sets if necessary.
+
+  Returns:
+    Paths to the training and test data files.
+  """
+
+  if not os.path.isdir(FLAGS.data_dir):
+    os.makedirs(FLAGS.data_dir)
+
+  training_data_path = os.path.join(FLAGS.data_dir,
+                                    os.path.basename(IRIS_TRAINING_DATA_URL))
+  if not os.path.isfile(training_data_path):
+    train_file = open(training_data_path, "wt")
+    urllib.request.urlretrieve(IRIS_TRAINING_DATA_URL, train_file.name)
+    train_file.close()
+
+    print("Training data are downloaded to %s" % train_file.name)
+
+  test_data_path = os.path.join(FLAGS.data_dir,
+                                os.path.basename(IRIS_TEST_DATA_URL))
+  if not os.path.isfile(test_data_path):
+    test_file = open(test_data_path, "wt")
+    urllib.request.urlretrieve(IRIS_TEST_DATA_URL, test_file.name)
+    test_file.close()
+
+    print("Test data are downloaded to %s" % test_file.name)
+
+  return training_data_path, test_data_path
+
+
+def main(_):
+  training_data_path, test_data_path = maybe_download_data()
+
+  # Load datasets.
+  training_set = tf.contrib.learn.datasets.base.load_csv_with_header(
+      filename=training_data_path,
+      target_dtype=np.int,
+      features_dtype=np.float32)
+  test_set = tf.contrib.learn.datasets.base.load_csv_with_header(
+      filename=test_data_path, target_dtype=np.int, features_dtype=np.float32)
+
+  # Specify that all features have real-value data
+  feature_columns = [tf.contrib.layers.real_valued_column("", dimension=4)]
+
+  # Build 3 layer DNN with 10, 20, 10 units respectively.
+  model_dir = FLAGS.model_dir or tempfile.mkdtemp(prefix="debug_tflearn_iris_")
+
+  classifier = tf.contrib.learn.DNNClassifier(
+      feature_columns=feature_columns,
+      hidden_units=[10, 20, 10],
+      n_classes=3,
+      model_dir=model_dir)
+
+  monitors = [tf_debug.LocalCLIDebugHook()] if FLAGS.debug else None
+
+  # Fit model.
+  classifier.fit(x=training_set.data,
+                 y=training_set.target,
+                 steps=FLAGS.train_steps,
+                 monitors=monitors)
+
+  # Evaluate accuracy.
+  accuracy_score = classifier.evaluate(
+      x=test_set.data, y=test_set.target)["accuracy"]
+  # TODO(cais): Add debug monitor for evaluate()
+
+  print("After training %d steps, Accuracy = %f" %
+        (FLAGS.train_steps, accuracy_score))
+
+
+if __name__ == "__main__":
+  tf.app.run()
diff --git a/tensorflow/python/debug/wrappers/hooks.py b/tensorflow/python/debug/wrappers/hooks.py
new file mode 100644
index 0000000000..14123d95f8
--- /dev/null
+++ b/tensorflow/python/debug/wrappers/hooks.py
@@ -0,0 +1,91 @@
+# Copyright 2016 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""tfdbg CLI as SessionRunHook."""
+
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+from tensorflow.core.protobuf import config_pb2
+from tensorflow.python.debug import debug_utils
+from tensorflow.python.debug.wrappers import framework
+from tensorflow.python.debug.wrappers import local_cli_wrapper
+from tensorflow.python.training import session_run_hook
+
+
+class LocalCLIDebugHook(session_run_hook.SessionRunHook,
+                        local_cli_wrapper.LocalCLIDebugWrapperSession):
+  """Command-line-interface debugger hook.
+
+  Can be used as a monitor/hook for tf.train.MonitoredSession.
+  """
+
+  def __init__(self):
+    """Create a local debugger command-line interface (CLI) hook."""
+
+    self._wrapper_initialized = False
+
+  def begin(self):
+    pass
+
+  def before_run(self, run_context):
+    if not self._wrapper_initialized:
+      local_cli_wrapper.LocalCLIDebugWrapperSession.__init__(
+          self, run_context.session)
+      self._wrapper_initialized = True
+
+    # Increment run call counter.
+    self._run_call_count += 1
+
+    # Adapt run_context to an instance of OnRunStartRequest for invoking
+    # superclass on_run_start().
+    on_run_start_request = framework.OnRunStartRequest(
+        run_context.original_args.fetches, run_context.original_args.feed_dict,
+        None, None, self._run_call_count)
+
+    on_run_start_response = self.on_run_start(on_run_start_request)
+    self._performed_action = on_run_start_response.action
+
+    run_args = session_run_hook.SessionRunArgs(
+        None, feed_dict=None, options=config_pb2.RunOptions())
+    if self._performed_action == framework.OnRunStartAction.DEBUG_RUN:
+      self._decorate_options_for_debug(run_args.options,
+                                       run_context.session.graph)
+    elif self._performed_action == framework.OnRunStartAction.INVOKE_STEPPER:
+      raise NotImplementedError(
+          "OnRunStartAction INVOKE_STEPPER has not been implemented.")
+
+    return run_args
+
+  def after_run(self, run_context, run_values):
+    # Adapt run_context and run_values to OnRunEndRequest and invoke superclass
+    # on_run_end()
+    if self._performed_action == framework.OnRunStartAction.DEBUG_RUN:
+      on_run_end_request = framework.OnRunEndRequest(self._performed_action,
+                                                     run_values.run_metadata)
+
+      self.on_run_end(on_run_end_request)
+
+  def _decorate_options_for_debug(self, options, graph):
+    """Modify RunOptions.debug_tensor_watch_opts for debugging.
+
+    Args:
+      options: (config_pb2.RunOptions) The RunOptions instance to be modified.
+      graph: A TensorFlow Graph object.
+    """
+
+    debug_utils.watch_graph(
+        options, graph, debug_urls=self._get_run_debug_urls())
+    options.output_partition_graphs = True
diff --git a/tensorflow/python/debug/wrappers/local_cli_wrapper.py b/tensorflow/python/debug/wrappers/local_cli_wrapper.py
index a5974afde8..74e3236e7d 100644
--- a/tensorflow/python/debug/wrappers/local_cli_wrapper.py
+++ b/tensorflow/python/debug/wrappers/local_cli_wrapper.py
@@ -23,6 +23,8 @@ import shutil
 import sys
 import tempfile
 
+import six
+
 # Google-internal import(s).
 from tensorflow.python.debug import debug_data
 from tensorflow.python.debug.cli import analyzer_cli
@@ -30,6 +32,7 @@ from tensorflow.python.debug.cli import curses_ui
 from tensorflow.python.debug.cli import debugger_cli_common
 from tensorflow.python.debug.wrappers import framework
 from tensorflow.python.framework import ops
+from tensorflow.python.ops import variables
 
 
 _DUMP_ROOT_PREFIX = "tfdbg_"
@@ -180,7 +183,10 @@ class LocalCLIDebugWrapperSession(framework.BaseDebugWrapperSession):
     else:
       feed_dict_lines = []
       for feed_key in request.feed_dict:
-        feed_dict_lines.append(feed_key.name)
+        if isinstance(feed_key, six.string_types):
+          feed_dict_lines.append(feed_key)
+        else:
+          feed_dict_lines.append(feed_key.name)
 
     # TODO(cais): Refactor into its own function.
     help_intro = [
@@ -457,11 +463,11 @@ class LocalCLIDebugWrapperSession(framework.BaseDebugWrapperSession):
     self._run_call_count = run_call_count
     self._run_description = "run #%d: " % self._run_call_count
 
-    if isinstance(fetches, ops.Tensor) or isinstance(fetches, ops.Operation):
+    if isinstance(fetches, (ops.Tensor, ops.Operation, variables.Variable)):
       self._run_description += "fetch: %s; " % fetches.name
     else:
       # Could be list, tuple, dict or namedtuple.
-      self._run_description += "%d fetch(es)" % len(fetches)
+      self._run_description += "%d fetch(es); " % len(fetches)
 
     if not feed_dict:
       self._run_description += "0 feeds"