diff --git a/python/paddle/trainer_config_helpers/layers.py b/python/paddle/trainer_config_helpers/layers.py
index a0432b3966b31da14232996d9c8c7857bddb11d9..56fca13d37242a091f09b35ca2e139071c644cad 100755
--- a/python/paddle/trainer_config_helpers/layers.py
+++ b/python/paddle/trainer_config_helpers/layers.py
@@ -5286,10 +5286,7 @@ def multi_binary_label_cross_entropy(input,
 def smooth_l1_cost(input, label, name=None, layer_attr=None):
     """
     This is a L1 loss but more smooth. It requires that the
-    size of input and label are equal.
-
-    More details can be found by referring to `Fast R-CNN
-    <https://arxiv.org/pdf/1504.08083v2.pdf>`_
+    size of input and label are equal. The formula is as follows,
 
     .. math::
 
@@ -5305,6 +5302,9 @@ def smooth_l1_cost(input, label, name=None, layer_attr=None):
         |x|-0.5& \text{otherwise}
         \end{cases}
 
+    More details can be found by referring to `Fast R-CNN
+    <https://arxiv.org/pdf/1504.08083v2.pdf>`_
+
     .. code-block:: python
 
        cost = smooth_l1_cost(input=input_layer,