diff --git a/paddle/operators/sigmoid_op.h b/paddle/operators/sigmoid_op.h
index 11ab923eb346c1f8de3a6bbebdfa874b6530004a..b01a9b3f23283471f8846325075719ba0e75ed35 100644
--- a/paddle/operators/sigmoid_op.h
+++ b/paddle/operators/sigmoid_op.h
@@ -37,7 +37,7 @@ class SigmoidKernel : public framework::OpKernel {
     auto Y = EigenVector<T>::Flatten(*output);
     auto place = context.GetEigenDevice<Place>();
 
-    Y.device(place) = 1.0 / (1.0 + (-1.0 * X).exp());
+    Y.device(place) = 1. / (1. + (-X).exp());
   }
 };
 
diff --git a/python/paddle/v2/framework/tests/gradient_checker.py b/python/paddle/v2/framework/tests/gradient_checker.py
index 2c92dfa43e7ee3abf4822268a8204d385c14bfc6..12f302fe251f60eac4e24525f3e019689095a7ad 100644
--- a/python/paddle/v2/framework/tests/gradient_checker.py
+++ b/python/paddle/v2/framework/tests/gradient_checker.py
@@ -188,10 +188,10 @@ class GradientChecker(unittest.TestCase):
 
         outputs = backward_op.outputs()
         out_names = [item for k in outputs for item in outputs[k]]
-        cpu_grads = self.get_grad(forward_op, backward_op, input_value,
-                                  out_names, core.CPUPlace())
-        gpu_grads = self.get_grad(forward_op, backward_op, input_value,
-                                  out_names, core.GPUPlace(0))
+        cpu_grads = self.__get_gradient(forward_op, backward_op, input_value,
+                                        out_names, core.CPUPlace())
+        gpu_grads = self.__get_gradient(forward_op, backward_op, input_value,
+                                        out_names, core.GPUPlace(0))
 
         for c_grad, g_grad, name in itertools.izip(cpu_grads, gpu_grads,
                                                    out_names):
@@ -277,8 +277,8 @@ class GradientChecker(unittest.TestCase):
         check_names = [grad_var_name(name) for name in inputs_to_check]
         for place in places:
             # get analytical gradients according to different device
-            analytic_grads = self.get_grad(forward_op, backward_op, input_vars,
-                                           check_names, place)
+            analytic_grads = self.__get_gradient(forward_op, backward_op,
+                                                 input_vars, check_names, place)
             self.__assert_is_close(numeric_grads, analytic_grads, check_names,
                                    max_relative_error,
                                    "Gradient Check On %s" % str(place))
diff --git a/python/paddle/v2/framework/tests/test_sigmoid_op.py b/python/paddle/v2/framework/tests/test_sigmoid_op.py
index c3bd79f5dc0ba4b087b681b3f8c241def3ba33c8..273c2e5ab1a84d12621fe9568c4cf22073b6aed4 100644
--- a/python/paddle/v2/framework/tests/test_sigmoid_op.py
+++ b/python/paddle/v2/framework/tests/test_sigmoid_op.py
@@ -14,14 +14,14 @@ class TestSigmoidOp(unittest.TestCase):
 
 
 class TestSigmoidGradOp(GradientChecker):
-    def test_compare_grad(self):
+    def test_grad(self):
         op = create_op("sigmoid")
-        inputs = {"X": np.random.random((11, 17)).astype("float32")}
+        inputs = {"X": np.random.uniform(0.1, 1, [11, 17]).astype("float32")}
         # compare gpu and cpu results for backward op.
-        # skip this test if only compiling CPU version.
+        # this test will be skiped if only compiling CPU version.
         self.compare_grad(op, inputs)
         # check gradients 
-        self.check_grad(op, inputs, set("X"), "Y")
+        self.check_grad(op, inputs, set("X"), "Y", max_relative_error=0.007)
 
 
 if __name__ == '__main__':