diff --git a/tensorflow/contrib/learn/python/learn/estimators/dnn_linear_combined.py b/tensorflow/contrib/learn/python/learn/estimators/dnn_linear_combined.py
index a5b1dc722f8..5ab5eec26a5 100644
--- a/tensorflow/contrib/learn/python/learn/estimators/dnn_linear_combined.py
+++ b/tensorflow/contrib/learn/python/learn/estimators/dnn_linear_combined.py
@@ -110,8 +110,7 @@ class _ComposableModel(object):
     grads = gradients.gradients(loss, my_vars)
     if self._gradient_clip_norm:
       grads, _ = clip_ops.clip_by_global_norm(grads, self._gradient_clip_norm)
-    self._optimizer = self._get_optimizer()
-    return [self._optimizer.apply_gradients(zip(grads, my_vars))]
+    return [self._get_optimizer().apply_gradients(zip(grads, my_vars))]
 
   def _get_feature_columns(self):
     if not self._feature_columns:
@@ -132,10 +131,12 @@ class _ComposableModel(object):
   def _get_optimizer(self):
     if (self._optimizer is None or isinstance(self._optimizer,
                                               six.string_types)):
-      self._optimizer = self._get_default_optimizer(self._optimizer)
+      optimizer = self._get_default_optimizer(self._optimizer)
     elif callable(self._optimizer):
-      self._optimizer = self._optimizer()
-    return self._optimizer
+      optimizer = self._optimizer()
+    else:
+      optimizer = self._optimizer
+    return optimizer
 
   def _get_default_optimizer(self, optimizer_name=None):
     raise NotImplementedError