diff --git a/text_classification/text_classification_cnn.py b/text_classification/text_classification_cnn.py
index 21995d5caa8692aed20d179f53d56d2a85a97f8e..87baccc57f8a8f924497f2be22497b57aaf1a399 100644
--- a/text_classification/text_classification_cnn.py
+++ b/text_classification/text_classification_cnn.py
@@ -38,7 +38,7 @@ def convolution_net(input_dim, class_dim=2, emb_dim=128, hid_dim=128):
 
     cost = paddle.layer.classification_cost(input=output, label=lbl)
 
-    return cost, output
+    return cost, output, lbl
 
 
 def train_cnn_model(num_pass):
@@ -57,7 +57,7 @@ def train_cnn_model(num_pass):
         lambda: paddle.dataset.imdb.test(word_dict), batch_size=100)
 
     # network config
-    [cost, _] = convolution_net(dict_dim, class_dim=class_dim)
+    [cost, output, label] = convolution_net(dict_dim, class_dim=class_dim)
     # create parameters
     parameters = paddle.parameters.create(cost)
     # create optimizer
@@ -66,6 +66,9 @@ def train_cnn_model(num_pass):
         regularization=paddle.optimizer.L2Regularization(rate=8e-4),
         model_average=paddle.optimizer.ModelAverage(average_window=0.5))
 
+    # add auc evaluator
+    paddle.evaluator.auc(input=output, label=label)
+
     # create trainer
     trainer = paddle.trainer.SGD(
         cost=cost, parameters=parameters, update_equation=adam_optimizer)
@@ -104,7 +107,7 @@ def cnn_infer(file_name):
     dict_dim = len(word_dict)
     class_dim = 2
 
-    [_, output] = convolution_net(dict_dim, class_dim=class_dim)
+    [_, output, _] = convolution_net(dict_dim, class_dim=class_dim)
     parameters = paddle.parameters.Parameters.from_tar(gzip.open(file_name))
 
     infer_data = []
diff --git a/text_classification/text_classification_dnn.py b/text_classification/text_classification_dnn.py
index 192deaa893388b00e9cbd9dbca7ce2d45151a9e2..af5bb67e9d32cdd8b292fb53064a0747eaf042d0 100644
--- a/text_classification/text_classification_dnn.py
+++ b/text_classification/text_classification_dnn.py
@@ -54,7 +54,7 @@ def fc_net(input_dim, class_dim=2, emb_dim=256):
 
     cost = paddle.layer.classification_cost(input=output, label=lbl)
 
-    return cost, output
+    return cost, output, lbl
 
 
 def train_dnn_model(num_pass):
@@ -73,7 +73,8 @@ def train_dnn_model(num_pass):
         lambda: paddle.dataset.imdb.test(word_dict), batch_size=100)
 
     # network config
-    [cost, _] = fc_net(dict_dim, class_dim=class_dim)
+    [cost, output, label] = fc_net(dict_dim, class_dim=class_dim)
+
     # create parameters
     parameters = paddle.parameters.create(cost)
     # create optimizer
@@ -82,6 +83,9 @@ def train_dnn_model(num_pass):
         regularization=paddle.optimizer.L2Regularization(rate=8e-4),
         model_average=paddle.optimizer.ModelAverage(average_window=0.5))
 
+    # add auc evaluator
+    paddle.evaluator.auc(input=output, label=label)
+
     # create trainer
     trainer = paddle.trainer.SGD(
         cost=cost, parameters=parameters, update_equation=adam_optimizer)
@@ -120,7 +124,7 @@ def dnn_infer(file_name):
     dict_dim = len(word_dict)
     class_dim = 2
 
-    [_, output] = fc_net(dict_dim, class_dim=class_dim)
+    [_, output, _] = fc_net(dict_dim, class_dim=class_dim)
     parameters = paddle.parameters.Parameters.from_tar(gzip.open(file_name))
 
     infer_data = []