diff --git a/fleetrec/models/ctr_dnn/data_generator.py b/fleetrec/models/ctr_dnn/data_generator.py
new file mode 100644
index 0000000000000000000000000000000000000000..3225cbd1b358b528b551ac21f62e0a56ae4bc3ab
--- /dev/null
+++ b/fleetrec/models/ctr_dnn/data_generator.py
@@ -0,0 +1,65 @@
+#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from __future__ import print_function
+from abc import ABC
+
+from fleetrec.reader.reader import Reader
+from fleetrec.utils import envs
+
+
+class TrainReader(Reader):
+    def init(self):
+        self.cont_min_ = [0, -3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
+        self.cont_max_ = [20, 600, 100, 50, 64000, 500, 100, 50, 500, 10, 10, 10, 50]
+        self.cont_diff_ = [20, 603, 100, 50, 64000, 500, 100, 50, 500, 10, 10, 10, 50]
+        self.hash_dim_ = envs.get_global_env("hyper_parameters.sparse_feature_number", None, "train.model")
+        self.continuous_range_ = range(1, 14)
+        self.categorical_range_ = range(14, 40)
+
+    def generate_sample(self, line):
+        """
+        Read the data line by line and process it as a dictionary
+        """
+
+        def reader():
+            """
+            This function needs to be implemented by the user, based on data format
+            """
+            features = line.rstrip('\n').split('\t')
+
+            dense_feature = []
+            sparse_feature = []
+            for idx in self.continuous_range_:
+                if features[idx] == "":
+                    dense_feature.append(0.0)
+                else:
+                    dense_feature.append(
+                        (float(features[idx]) - self.cont_min_[idx - 1]) /
+                        self.cont_diff_[idx - 1])
+
+            for idx in self.categorical_range_:
+                sparse_feature.append(
+                    [hash(str(idx) + features[idx]) % self.hash_dim_])
+            label = [int(features[0])]
+            feature_name = ["dense_input"]
+            for idx in self.categorical_range_:
+                feature_name.append("C" + str(idx - 13))
+            feature_name.append("label")
+            yield zip(feature_name, [dense_feature] + sparse_feature + [label])
+
+        return reader
+
+
+class EvaluateReader(Reader, ABC):
+    pass
diff --git a/fleetrec/reader/reader.py b/fleetrec/reader/reader.py
new file mode 100644
index 0000000000000000000000000000000000000000..8e696747f99f63ae2f3323ea851323f1c61d50f1
--- /dev/null
+++ b/fleetrec/reader/reader.py
@@ -0,0 +1,30 @@
+#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from __future__ import print_function
+
+import abc
+
+import paddle.fluid.incubate.data_generator as dg
+
+
+class Reader(dg.MultiSlotDataGenerator):
+    __metaclass__ = abc.ABCMeta
+
+    @abc.abstractmethod
+    def init(self):
+        pass
+
+    @abc.abstractmethod
+    def generate_sample(self, line):
+        pass
diff --git a/fleetrec/reader/reader_instance.py b/fleetrec/reader/reader_instance.py
new file mode 100644
index 0000000000000000000000000000000000000000..2b771794c23d0eaad22f436358ba72027f250e1a
--- /dev/null
+++ b/fleetrec/reader/reader_instance.py
@@ -0,0 +1,31 @@
+#   Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from __future__ import print_function
+import sys
+
+from fleetrec.utils.envs import lazy_instance
+
+if len(sys.argv) != 3:
+    raise ValueError("reader only accept two argument: initialized reader class name and TRAIN/EVALUATE")
+
+reader_package = sys.argv[1]
+
+if sys.argv[2] == "TRAIN":
+    reader_name = "TrainReader"
+else:
+    reader_name = "EvaluateReader"
+
+reader_class = lazy_instance(reader_package, reader_name)
+reader = reader_class()
+reader.run_from_stdin()
diff --git a/fleetrec/trainer/transpiler_trainer.py b/fleetrec/trainer/transpiler_trainer.py
index f7fa8e679d27b9480479e5e7f9b4200f76855151..fa568851986e02585f3f556c9f204247116ee1db 100644
--- a/fleetrec/trainer/transpiler_trainer.py
+++ b/fleetrec/trainer/transpiler_trainer.py
@@ -21,8 +21,8 @@ import paddle.fluid as fluid
 
 from paddle.fluid.incubate.fleet.parameter_server.distribute_transpiler import fleet
 
-from .trainer import Trainer
-from ..utils import envs
+from fleetrec.trainer import Trainer
+from fleetrec.utils import envs
 
 
 class TranspileTrainer(Trainer):
@@ -113,9 +113,8 @@ class TranspileTrainer(Trainer):
 
     def instance(self, context):
         models = envs.get_global_env("train.model.models")
-        model_package = __import__(models, globals(), locals(), models.split("."))
-        train_model = getattr(model_package, 'Train')
-        self.model = train_model(None)
+        model_class = envs.lazy_instance(models, "TrainNet")
+        self.model = model_class(None)
         context['status'] = 'init_pass'
 
     def init(self, context):
diff --git a/fleetrec/utils/envs.py b/fleetrec/utils/envs.py
index 95cf444741c60a88d2d50616b6d9b7097e80d0c9..2cd55dd47d251ee9b9f0b3229595d5284369da6d 100644
--- a/fleetrec/utils/envs.py
+++ b/fleetrec/utils/envs.py
@@ -81,3 +81,11 @@ def pretty_print_envs(envs, header=None):
 
     _str = "\n{}\n".format(draws)
     return _str
+
+
+def lazy_instance(package, class_name):
+    models = get_global_env("train.model.models")
+    model_package = __import__(package, globals(), locals(), package.split("."))
+    instance = getattr(model_package, class_name)
+    return instance
+