diff --git a/tensorflow/contrib/layers/python/layers/feature_column_ops.py b/tensorflow/contrib/layers/python/layers/feature_column_ops.py
index 823a38f30d0..f343b68f7c0 100644
--- a/tensorflow/contrib/layers/python/layers/feature_column_ops.py
+++ b/tensorflow/contrib/layers/python/layers/feature_column_ops.py
@@ -98,9 +98,13 @@ def input_from_feature_columns(columns_to_tensors,
                                     [ops.GraphKeys.VARIABLES]))
 
     for column in sorted(set(feature_columns), key=lambda x: x.key):
-      transformed_tensor = transformer.transform(column)
-      output_tensors.append(column.to_dnn_input_layer(
-          transformed_tensor, weight_collections, trainable))
+      try:
+        transformed_tensor = transformer.transform(column)
+        output_tensors.append(column.to_dnn_input_layer(
+            transformed_tensor, weight_collections, trainable))
+      except ValueError as e:
+        raise ValueError('Error creating input layer for column: {}.\n'
+                         '{}'.format(column.name, e))
     return array_ops.concat(1, output_tensors)
 
 
@@ -174,11 +178,15 @@ def weighted_sum_from_feature_columns(columns_to_tensors,
     column_to_variable = dict()
     transformer = _Transformer(columns_to_tensors)
     for column in sorted(set(feature_columns), key=lambda x: x.key):
-      transformed_tensor = transformer.transform(column)
-      predictions, variable = column.to_weighted_sum(transformed_tensor,
-                                                     num_outputs,
-                                                     weight_collections,
-                                                     trainable)
+      try:
+        transformed_tensor = transformer.transform(column)
+        predictions, variable = column.to_weighted_sum(transformed_tensor,
+                                                       num_outputs,
+                                                       weight_collections,
+                                                       trainable)
+      except ValueError as e:
+        raise ValueError('Error creating weighted sum for column: {}.\n'
+                         '{}'.format(column.name, e))
       output_tensors.append(predictions)
       column_to_variable[column] = variable
       _log_variable(variable)
@@ -305,7 +313,10 @@ def check_feature_columns(feature_columns):
   for f in feature_columns:
     key = f.key
     if key in seen_keys:
-      raise ValueError('Duplicate feature column key found: %s' % key)
+      raise ValueError('Duplicate feature column key found for column: {}. '
+                       'This usually means that the column is almost identical '
+                       'to another column, and one must be discarded.'.format(
+                           f.name))
     seen_keys.add(key)
 
 
diff --git a/tensorflow/contrib/layers/python/layers/feature_column_ops_test.py b/tensorflow/contrib/layers/python/layers/feature_column_ops_test.py
index 1d0f45357ed..33aa3c8b091 100644
--- a/tensorflow/contrib/layers/python/layers/feature_column_ops_test.py
+++ b/tensorflow/contrib/layers/python/layers/feature_column_ops_test.py
@@ -341,9 +341,12 @@ class InputLayerTest(tf.test.TestCase):
 
     # Makes sure that trying to use different initializers with the same
     # embedding column explicitly fails.
-    with self.assertRaises(ValueError):
-      tf.contrib.layers.input_from_feature_columns(
-          features, [embedded_sparse, embedded_sparse_alternate])
+    with self.test_session():
+      with self.assertRaisesRegexp(
+          ValueError,
+          "Duplicate feature column key found for column: wire_embedding"):
+        tf.contrib.layers.input_from_feature_columns(
+            features, [embedded_sparse, embedded_sparse_alternate])
 
   def testSparseColumn(self):
     hashed_sparse = tf.contrib.layers.sparse_column_with_hash_bucket("wire", 10)
@@ -351,9 +354,11 @@ class InputLayerTest(tf.test.TestCase):
                                   indices=[[0, 0], [1, 0], [1, 1]],
                                   shape=[2, 2])
     features = {"wire": wire_tensor}
-    with self.assertRaises(ValueError):
-      tf.initialize_all_variables().run()
-      tf.contrib.layers.input_layer(features, [hashed_sparse])
+    with self.test_session():
+      with self.assertRaisesRegexp(
+          ValueError, "Error creating input layer for column: wire"):
+        tf.initialize_all_variables().run()
+        tf.contrib.layers.input_from_feature_columns(features, [hashed_sparse])
 
   def testCrossedColumn(self):
     a = tf.contrib.layers.sparse_column_with_hash_bucket("aaa",
@@ -366,9 +371,11 @@ class InputLayerTest(tf.test.TestCase):
                                   indices=[[0, 0], [1, 0], [1, 1]],
                                   shape=[2, 2])
     features = {"aaa": wire_tensor, "bbb": wire_tensor}
-    with self.assertRaises(ValueError):
-      tf.initialize_all_variables().run()
-      tf.contrib.layers.input_layer(features, [crossed])
+    with self.test_session():
+      with self.assertRaisesRegexp(
+          ValueError, "Error creating input layer for column: aaa_X_bbb"):
+        tf.initialize_all_variables().run()
+        tf.contrib.layers.input_from_feature_columns(features, [crossed])
 
   def testAllColumns(self):
     real_valued = tf.contrib.layers.real_valued_column("income", 3)
@@ -477,10 +484,13 @@ class WeightedSumTest(tf.test.TestCase):
                                   shape=[2, 2])
     features = {"wire": wire_tensor}
     embeded_sparse = tf.contrib.layers.embedding_column(hashed_sparse, 10)
-    with self.assertRaises(ValueError):
-      tf.initialize_all_variables().run()
-      tf.contrib.layers.weighted_sum_from_feature_columns(features,
-                                                          [embeded_sparse])
+    with self.test_session():
+      with self.assertRaisesRegexp(
+          ValueError, "Error creating weighted sum for column: wire_embedding"):
+        tf.initialize_all_variables().run()
+        tf.contrib.layers.weighted_sum_from_feature_columns(features,
+                                                            [embeded_sparse],
+                                                            num_outputs=5)
 
   def testRealValuedColumnWithMultiDimensions(self):
     real_valued = tf.contrib.layers.real_valued_column("price", 2)