diff --git a/tensorflow/python/training/optimizer.py b/tensorflow/python/training/optimizer.py
index 059f59800ea..a9871291196 100644
--- a/tensorflow/python/training/optimizer.py
+++ b/tensorflow/python/training/optimizer.py
@@ -656,7 +656,7 @@ def apply_fn():
           with ops.name_scope("update_" + scope_name), ops.colocate_with(var):
             update_ops.append(processor.update_op(self, grad))
         if global_step is None:
-          apply_updates = self._finish(update_ops, sname+'-apply')
+          apply_updates = self._finish(update_ops, sname)
         else:
           with ops.control_dependencies([self._finish(update_ops, "update")]):
             with ops.colocate_with(global_step):
@@ -666,9 +666,9 @@ def apply_fn():
                 apply_updates = resource_variable_ops.assign_add_variable_op(
                     global_step.handle,
                     ops.convert_to_tensor(1, dtype=global_step.dtype),
-                    name=sname+'-apply')
+                    name=sname)
               else:
-                apply_updates = state_ops.assign_add(global_step, 1, name=sname+'-apply')
+                apply_updates = state_ops.assign_add(global_step, 1, name=sname)
 
         if not context.executing_eagerly():
           if isinstance(apply_updates, ops.Tensor):
@@ -772,12 +772,12 @@ def finish(self, update_ops):
         finish_updates = distribution.extended.update_non_slot(
             non_slot_devices, finish, args=(self, update_ops), group=False)
         if global_step is None:
-          apply_updates = distribution.group(finish_updates, name=sname+'-apply')
+          apply_updates = distribution.group(finish_updates, name=sname)
         else:
           with ops.control_dependencies(finish_updates):
             apply_updates = distribution.extended.update(
                 global_step, state_ops.assign_add, args=(1,),
-                kwargs={"name": sname+'-apply'})
+                kwargs={"name": sname})
 
         if not context.executing_eagerly():
           if isinstance(apply_updates, ops.Tensor):