diff --git a/fluid/ocr_recognition/ctc_train.py b/fluid/ocr_recognition/ctc_train.py
index 922f70f05722fa14d322b2061900092fd2adb0dd..4a68ebdd2e5fd0b50c9085965ce9ebed7829281c 100644
--- a/fluid/ocr_recognition/ctc_train.py
+++ b/fluid/ocr_recognition/ctc_train.py
@@ -81,17 +81,16 @@ def train(args, data_reader=dummy_reader):
                 sys.stdout.flush()
             batch_id += 1
 
-        if model_average != None:
-            model_average.apply(exe)
-        error_evaluator.reset(exe)
-        for data in test_reader():
-            exe.run(inference_program, feed=get_feeder_data(data, place))
-        _, test_seq_error = error_evaluator.eval(exe)
-        if model_average != None:
-            model_average.restore(exe)
+        with model_average.apply(exe):
+            error_evaluator.reset(exe)
+            for data in test_reader():
+                exe.run(inference_program, feed=get_feeder_data(data, place))
+            _, test_seq_error = error_evaluator.eval(exe)
+            if model_average != None:
+                model_average.restore(exe)
 
-        print "\nEnd pass[%d]; Test seq error: %s.\n" % (
-            pass_id, str(test_seq_error[0]))
+            print "\nEnd pass[%d]; Test seq error: %s.\n" % (
+                pass_id, str(test_seq_error[0]))
 
 def main():
     args = parser.parse_args()