diff --git a/fluid_onnx/ops.py b/fluid_onnx/ops.py
index d91315268..5516e3075 100644
--- a/fluid_onnx/ops.py
+++ b/fluid_onnx/ops.py
@@ -58,6 +58,11 @@
 
 
 def activation_ops(act_type, operator, block):
+    """ Convert common activations with type specified by 'act_type', including
+        'abs', 'ceil', 'exp', 'floor', 'log', 'reciprocal', 'relu', 'sigmoid',
+        'softplus', 'softsign', 'sqrt' and 'tanh'.
+    """
+
     inputs, _, outputs = op_io_info(operator)
     return make_node(
         act_type, inputs=inputs.values()[0], outputs=outputs.values()[0])
@@ -114,16 +119,32 @@ def batch_norm_op(operator, block):
                                                  bn_node)
 
 
-def cast_op():
-    pass
+def cast_op(operator, block):
+    inputs, attrs, outputs = op_io_info(operator)
+    return make_node(
+        'Cast',
+        inputs=inputs['X'],
+        outputs=outputs['Out'],
+        to=PADDLE_TO_ONNX_DTYPE[attrs['out_dtype']])
 
 
-def clip_op():
-    pass
+def clip_op(operator, block):
+    inputs, attrs, outputs = op_io_info(operator)
+    return make_node(
+        'Clip',
+        inputs=inputs['X'],
+        outputs=outputs['Out'],
+        min=attrs['min'],
+        max=attrs['max'])
 
 
-def concat_op():
-    pass
+def concat_op(operator, block):
+    inputs, attrs, outputs = op_io_info(operator)
+    return make_node(
+        'Concat',
+        inputs=inputs['X'],
+        outputs=outputs['Out'],
+        axis=attrs['axis'])
 
 
 def constant_op(var, scope):
@@ -156,8 +177,20 @@ def conv2d_op(operator, block):
     return conv2d
 
 
-def convtranspose_op():
-    pass
+def conv2d_transpose_op(operator, block):
+    inputs, attrs, outputs = op_io_info(operator)
+
+    kernel_shape = block.vars[inputs['Filter'][0]].shape
+    conv2d_transpose = make_node(
+        'ConvTranspose',
+        inputs=inputs['Input'] + inputs['Filter'],
+        outputs=outputs['Output'],
+        dilations=attrs['dilations'],
+        kernel_shape=kernel_shape[-2:],
+        strides=attrs['strides'],
+        group=1,
+        pads=attrs['paddings'] + attrs['paddings'])
+    return conv2d_transpose
 
 
 def depthtospace_op():
@@ -184,12 +217,19 @@ def dropout_op(operator, block):
 
 
 def elementwise_ops(op_type, operator, block):
+    """Convert elementwise operators From to ONNX. Supported elementwise 
+       'op_type' includes 'Add', 'Div', 'Mul', 'Pow' and 'Sub'. 
+    """
+
     inputs, attrs, outputs = op_io_info(operator)
+    rank_x = len(block.vars[inputs['X'][0]].shape)
+    rank_y = len(block.vars[inputs['Y'][0]].shape)
+    axis = rank_x - rank_y if attrs['axis'] == -1 else attrs['axis']
     return make_node(
         op_type,
         inputs=inputs['X'] + inputs['Y'],
         outputs=outputs['Out'],
-        axis=attrs['axis'],
+        axis=axis,
         broadcast=1)
 
 
@@ -260,8 +300,21 @@ def less_op():
     pass
 
 
-def log_op():
-    pass
+def binary_logical_ops(op_type, operator, block):
+    """Convert binary logical operators, i.e. 'And', 'Or' and 'Xor'.
+    """
+
+    inputs, _, outputs = op_io_info(operator)
+    return make_node(
+        op_type, inputs=inputs['X'] + inputs['Y'], outputs=outputs['Out'])
+
+
+def unary_logical_ops(op_type, operator, block):
+    """Convert unary logical operators, i.e. 'Not'.
+    """
+
+    inputs, _, outputs = op_io_info(operator)
+    return make_node(op_type, inputs=inputs['X'], outputs=outputs['Out'])
 
 
 def logsoftmax_op():
@@ -422,6 +475,11 @@ def randomuniformlike_op():
 
 
 def reduce_ops(op_type, operator, block):
+    """Convert reduce operators in Fluid to ONNX. 'op_type' specifies the 
+       target ONNX operator type, supporting 'Reduce{Max, Mean, Min, Sum}'
+       right now.
+    """
+
     inputs, attrs, outputs = op_io_info(operator)
     rank = len(block.vars[inputs['X'][0]].shape)
     dim = attrs['dim']
@@ -549,19 +607,17 @@ def xor_op():
 node_maker = {
     # Paddle op name : (ONNX op name, modifier)
     'abs': partial(activation_ops, 'Abs'),
-    # '': 'And', # ?
     # 'ArgMax', NEEDS ATTENTION.
     # 'ArgMin', NEEDS ATTENTION.
     'batch_norm': batch_norm_op,
-    'cast': ('Cast', cast_op),
+    'cast': cast_op,
     'ceil': partial(activation_ops, 'Ceil'),
-    'clip': ('Clip', clip_op),
-    'concat': ('Concat', concat_op),
+    'clip': clip_op,
+    'concat': concat_op,
     'constant': constant_op,
     'conv2d': conv2d_op,
-
     # Need to continue the mapping below.
-    '': 'ConvTranspose',
+    'conv2d_transpose': conv2d_transpose_op,
     '': 'DepthToSpace',
     'depthwise_conv2d': conv2d_op,
     'dropout': dropout_op,
@@ -588,6 +644,10 @@ def xor_op():
     '': 'LeakyRelu',
     '': 'Less',
     'log': partial(activation_ops, 'Log'),
+    'logical_and': partial(binary_logical_ops, 'And'),
+    'logical_or': partial(binary_logical_ops, 'Or'),
+    'logical_not': partial(unary_logical_ops, 'Not'),
+    'logical_xor': partial(binary_logical_ops, 'Xor'),
     ',': 'LogSoftmax',
     '': 'LpNormalization',
     '': 'LpPool',
@@ -599,8 +659,6 @@ def xor_op():
     '': 'Min',
     'mul': mul_op,
     ',': 'Neg',
-    '': 'Not',
-    '': 'Or',
     '': 'PRelu',
     '': 'Pad',
     'pool2d': pool2d_op,
@@ -640,7 +698,6 @@ def xor_op():
     '': 'TopK',
     '': 'Transpose',
     # 'Unsqueeze', NEEDS ATTENTION.
-    '': 'Xor',
     # 'experimental ATen'
     # ',': 'experimental Affine'
     # 'experimental ConstantFill'
diff --git a/fluid_onnx/variables.py b/fluid_onnx/variables.py
index 4db7f3aa8..47af770ce 100644
--- a/fluid_onnx/variables.py
+++ b/fluid_onnx/variables.py
@@ -36,7 +36,7 @@ def paddle_onnx_shape(paddle_shape):
 
 PADDLE_TO_ONNX_DTYPE = {
     core.VarDesc.VarType.FP32: onnx_pb2.TensorProto.FLOAT,
-    core.VarDesc.VarType.FP64: onnx_pb2.TensorProto.FLOAT16,
+    core.VarDesc.VarType.FP64: onnx_pb2.TensorProto.DOUBLE,
     # '': onnx_pb2.TensorProto.DOUBLE,
     core.VarDesc.VarType.INT32: onnx_pb2.TensorProto.INT32,
     core.VarDesc.VarType.INT16: onnx_pb2.TensorProto.INT16,
diff --git a/tests/op_test.py b/tests/op_test.py
index 22bae505b..42b973701 100644
--- a/tests/op_test.py
+++ b/tests/op_test.py
@@ -77,7 +77,7 @@ def create_var(block, name, np_list, var_proto):
 
         persistable = True if name in persistable_list else False
         return block.create_var(
-            dtype="float32",
+            dtype='float32',
             shape=shape,
             persistable=persistable,
             lod_level=lod_level,
@@ -263,9 +263,17 @@ def eval_onnx_node(self):
         onnx_graph = make_graph(node_list, self.op_type, inputs, outputs)
         onnx_model = make_model(onnx_graph, producer_name='unittest')
 
+        # Expand input dictionary if there are tensor arrays
+        input_map = {}
+        for v in self.inputs:
+            if isinstance(self.inputs[v], list):
+                input_map.update(self.inputs[v])
+            else:
+                input_map[v] = self.inputs[v]
+
         # Run the Caffe2Backend with the ONNX model.
         rep = Caffe2Backend.prepare(onnx_model, device='CPU')
-        in_vals = [self.inputs[input.name] for input in inputs]
+        in_vals = [input_map[input.name] for input in inputs]
         outs = rep.run(in_vals)
 
         return outs
diff --git a/tests/test_activation_ops.py b/tests/test_activation_ops.py
index 2b0f55de2..b1310c2bd 100644
--- a/tests/test_activation_ops.py
+++ b/tests/test_activation_ops.py
@@ -19,9 +19,9 @@
 
 class TestAbsOp(OpTest):
     def setUp(self):
-        X = np.random.random((13, 15)).astype("float32")
+        X = np.random.random((13, 15)).astype('float32')
         self.inputs = {'X': X}
-        self.outputs = {'Out': np.zeros((1, 1)).astype("float32")}
+        self.outputs = {'Out': np.zeros((1, 1)).astype('float32')}
         self.init_op_type()
 
     def init_op_type(self):
diff --git a/tests/test_cast_op.py b/tests/test_cast_op.py
new file mode 100644
index 000000000..67903dd9c
--- /dev/null
+++ b/tests/test_cast_op.py
@@ -0,0 +1,37 @@
+#   Copyright (c) 2018 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.
+
+import unittest
+import numpy as np
+import paddle.fluid.core as core
+from op_test import OpTest
+
+
+class TestCastOp(OpTest):
+    def setUp(self):
+        input = np.random.random((10, 10))
+        self.inputs = {'X': input.astype('float32')}
+        self.outputs = {'Out': input.astype('float64')}
+        self.attrs = {
+            'in_dtype': int(core.VarDesc.VarType.FP32),
+            'out_dtype': int(core.VarDesc.VarType.FP64)
+        }
+        self.op_type = 'cast'
+
+    def test_check_output(self):
+        self.check_output()
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/tests/test_clip_op.py b/tests/test_clip_op.py
new file mode 100644
index 000000000..d39c8af43
--- /dev/null
+++ b/tests/test_clip_op.py
@@ -0,0 +1,33 @@
+# Copyright (c) 2018 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.
+
+import unittest
+import numpy as np
+from op_test import OpTest
+
+
+class TestClipOp(OpTest):
+    def setUp(self):
+        input = np.random.random((4, 5, 6)).astype('float32')
+        self.op_type = 'clip'
+        self.inputs = {'X': input}
+        self.attrs = {'min': 0.2, 'max': 0.8}
+        self.outputs = {'Out': np.zeros((1, 1)).astype('float32')}
+
+    def test_check_output(self):
+        self.check_output()
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/tests/test_concat_op.py b/tests/test_concat_op.py
new file mode 100644
index 000000000..40943962f
--- /dev/null
+++ b/tests/test_concat_op.py
@@ -0,0 +1,47 @@
+#   Copyright (c) 2018 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.
+
+import unittest
+import numpy as np
+from op_test import OpTest
+
+
+class TestConcatOp(OpTest):
+    def setUp(self):
+        self.op_type = 'concat'
+        self.init_test_data()
+        self.inputs = {'X': [('x0', self.x0), ('x1', self.x1), ('x2', self.x2)]}
+        self.attrs = {'axis': self.axis}
+        self.outputs = {'Out': np.zeros((1, 1)).astype('float32')}
+
+    def test_check_output(self):
+        self.check_output()
+
+    def init_test_data(self):
+        self.x0 = np.random.random((2, 1, 4, 5)).astype('float32')
+        self.x1 = np.random.random((2, 2, 4, 5)).astype('float32')
+        self.x2 = np.random.random((2, 3, 4, 5)).astype('float32')
+        self.axis = 1
+
+
+class TestConcatOp2(OpTest):
+    def init_test_data(self):
+        self.x0 = np.random.random((2, 3, 4, 5)).astype('float32')
+        self.x1 = np.random.random((2, 3, 4, 5)).astype('float32')
+        self.x2 = np.random.random((2, 3, 4, 5)).astype('float32')
+        self.axis = 2
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/tests/test_conv2d_op.py b/tests/test_conv2d_op.py
index 40aeed9bf..e5d797240 100644
--- a/tests/test_conv2d_op.py
+++ b/tests/test_conv2d_op.py
@@ -57,7 +57,7 @@ def setUp(self):
         self.outputs = {'Output': output}
 
     def init_conv_type(self):
-        self.op_type = "conv2d"
+        self.op_type = 'conv2d'
 
     def test_check_output(self):
         self.check_output(decimal=5)
@@ -65,7 +65,7 @@ def test_check_output(self):
 
 class TestDepthwiseConv2dOp(TestConv2dOp):
     def init_conv_type(self):
-        self.op_type = "depthwise_conv2d"
+        self.op_type = 'depthwise_conv2d'
 
 
 if __name__ == '__main__':
diff --git a/tests/test_conv2d_transpose_op.py b/tests/test_conv2d_transpose_op.py
new file mode 100644
index 000000000..2fb7f39ab
--- /dev/null
+++ b/tests/test_conv2d_transpose_op.py
@@ -0,0 +1,56 @@
+#   Copyright (c) 2018 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.
+
+import unittest
+import numpy as np
+from op_test import OpTest
+
+
+class TestConv2dTransposeOp(OpTest):
+    def setUp(self):
+        self.use_cudnn = False
+        self.init_op_type()
+        self.init_test_case()
+
+        input_ = np.random.random(self.input_size).astype('float32')
+        filter_ = np.random.random(self.filter_size).astype('float32')
+
+        self.inputs = {'Input': input_, 'Filter': filter_}
+        self.attrs = {
+            'strides': self.stride,
+            'paddings': self.pad,
+            'dilations': self.dilations,
+            'use_cudnn': self.use_cudnn,
+            'data_format': 'AnyLayout'
+        }
+
+        self.outputs = {'Output': np.zeros((1, 1))}
+
+    def init_test_case(self):
+        self.pad = [0, 0]
+        self.stride = [1, 1]
+        self.dilations = [1, 1]
+        self.input_size = [2, 3, 5, 5]  # NCHW
+        f_c = self.input_size[1]
+        self.filter_size = [f_c, 6, 3, 3]
+
+    def init_op_type(self):
+        self.op_type = 'conv2d_transpose'
+
+    def test_check_output(self):
+        self.check_output()
+
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/tests/test_dropout_op.py b/tests/test_dropout_op.py
index d503b7e60..397c2e482 100644
--- a/tests/test_dropout_op.py
+++ b/tests/test_dropout_op.py
@@ -19,8 +19,8 @@
 
 class TestDropoutOp(OpTest):
     def setUp(self):
-        self.op_type = "dropout"
-        self.inputs = {'X': np.random.random((32, 64, 2)).astype("float32")}
+        self.op_type = 'dropout'
+        self.inputs = {'X': np.random.random((32, 64, 2)).astype('float32')}
         self.attrs = {'dropout_prob': 0.8, 'is_test': True}
         self.outputs = {'Out': np.zeros((1, 1))}
 
diff --git a/tests/test_elementwise_ops.py b/tests/test_elementwise_ops.py
index 76c189391..37b01932a 100644
--- a/tests/test_elementwise_ops.py
+++ b/tests/test_elementwise_ops.py
@@ -19,41 +19,47 @@
 
 class TestElementwiseAddOp(OpTest):
     def setUp(self):
-        self.init_op_type()
-        self.attrs = {"axis": 1}
+        self.attrs = {'axis': 2}
+        self.init()
 
         self.inputs = {
-            'X': np.random.random((4, 2)).astype(np.float32),
-            'Y': np.random.random((2, )).astype(np.float32)
+            'X': np.random.random((2, 3, 4, 5)).astype(np.float32),
+            'Y': np.random.random((4, 5)).astype(np.float32)
         }
 
         self.outputs = {'Out': np.zeros((1, 1))}
 
-    def init_op_type(self):
-        self.op_type = "elementwise_add"
+    def init(self):
+        self.op_type = 'elementwise_add'
 
     def test_check_output(self):
         self.check_output()
 
 
+class TestElementwiseAddOpNegAxis(OpTest):
+    def init(self):
+        self.op_type = 'elementwise_add'
+        self.attrs = {'axis': -1}
+
+
 class TestElementwiseSubOp(TestElementwiseAddOp):
-    def init_op_type(self):
-        self.op_type = "elementwise_sub"
+    def init(self):
+        self.op_type = 'elementwise_sub'
 
 
 class TestElementwiseMulOp(TestElementwiseAddOp):
-    def init_op_type(self):
-        self.op_type = "elementwise_mul"
+    def init(self):
+        self.op_type = 'elementwise_mul'
 
 
 class TestElementwiseDivOp(TestElementwiseAddOp):
-    def init_op_type(self):
-        self.op_type = "elementwise_div"
+    def init(self):
+        self.op_type = 'elementwise_div'
 
 
 class TestElementwisePowOp(TestElementwiseAddOp):
-    def init_op_type(self):
-        self.op_type = "elementwise_pow"
+    def init(self):
+        self.op_type = 'elementwise_pow'
 
 
 if __name__ == '__main__':
diff --git a/tests/test_logical_ops.py b/tests/test_logical_ops.py
new file mode 100644
index 000000000..1d7dfe60f
--- /dev/null
+++ b/tests/test_logical_ops.py
@@ -0,0 +1,49 @@
+#   Copyright (c) 2018 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.
+
+import op_test
+import unittest
+import numpy as np
+
+
+def create_test_class(op_type, callback, binary_op=True):
+    class Cls(op_test.OpTest):
+        def setUp(self):
+            a = np.random.choice(a=[True, False], size=(10, 7)).astype(bool)
+            if binary_op:
+                b = np.random.choice(a=[True, False], size=(10, 7)).astype(bool)
+                c = callback(a, b)
+            else:
+                c = callback(a)
+            self.outputs = {'Out': c}
+            self.op_type = op_type
+            if binary_op:
+                self.inputs = {'X': a, 'Y': b}
+            else:
+                self.inputs = {'X': a}
+
+        def test_output(self):
+            self.check_output()
+
+    Cls.__name__ = op_type
+    globals()[op_type] = Cls
+
+
+create_test_class('logical_and', lambda _a, _b: np.logical_and(_a, _b))
+create_test_class('logical_or', lambda _a, _b: np.logical_or(_a, _b))
+create_test_class('logical_not', lambda _a: np.logical_not(_a), False)
+create_test_class('logical_xor', lambda _a, _b: np.logical_xor(_a, _b))
+
+if __name__ == '__main__':
+    unittest.main()
diff --git a/tests/test_mul_op.py b/tests/test_mul_op.py
index 0b2e73f30..e36a8a6bd 100644
--- a/tests/test_mul_op.py
+++ b/tests/test_mul_op.py
@@ -19,10 +19,10 @@
 
 class TestMulOp(OpTest):
     def setUp(self):
-        self.op_type = "mul"
+        self.op_type = 'mul'
         self.inputs = {
-            'X': np.random.random((15, 4, 12, 10)).astype("float32"),
-            'Y': np.random.random((4, 30, 8, 2, 9)).astype("float32")
+            'X': np.random.random((15, 4, 12, 10)).astype('float32'),
+            'Y': np.random.random((4, 30, 8, 2, 9)).astype('float32')
         }
         self.attrs = {'x_num_col_dims': 2, 'y_num_col_dims': 2}
         result = np.dot(self.inputs['X'].reshape(15 * 4, 12 * 10),
diff --git a/tests/test_pool2d_op.py b/tests/test_pool2d_op.py
index bff216d9e..2ab6c71b6 100644
--- a/tests/test_pool2d_op.py
+++ b/tests/test_pool2d_op.py
@@ -19,7 +19,7 @@
 
 class TestPool2dOp(OpTest):
     def setUp(self):
-        self.op_type = "pool2d"
+        self.op_type = 'pool2d'
         self.use_cudnn = False
         self.use_mkldnn = False
         self.dtype = np.float32
@@ -79,7 +79,7 @@ def init_test_case(self):
         self.paddings = [0, 0]
 
     def init_pool_type(self):
-        self.pool_type = "avg"
+        self.pool_type = 'avg'
 
     def init_global_pool(self):
         self.global_pool = False
@@ -93,7 +93,7 @@ def init_test_case(self):
         self.paddings = [1, 1]
 
     def init_pool_type(self):
-        self.pool_type = "avg"
+        self.pool_type = 'avg'
 
     def init_global_pool(self):
         self.global_pool = False
@@ -101,17 +101,17 @@ def init_global_pool(self):
 
 class TestPool2dOp3(TestPool2dOp):
     def init_pool_type(self):
-        self.pool_type = "max"
+        self.pool_type = 'max'
 
 
 class TestPool2dOp4(TestPool2dOp1):
     def init_pool_type(self):
-        self.pool_type = "max"
+        self.pool_type = 'max'
 
 
 class TestPool2dOp5(TestPool2dOp2):
     def init_pool_type(self):
-        self.pool_type = "max"
+        self.pool_type = 'max'
 
 
 if __name__ == '__main__':
diff --git a/tests/test_reduce_ops.py b/tests/test_reduce_ops.py
index 7f43f3a6a..5dba8321e 100644
--- a/tests/test_reduce_ops.py
+++ b/tests/test_reduce_ops.py
@@ -22,7 +22,7 @@ def setUp(self):
         self.init_op_type()
         self.init_keep_dim()
         self.init_reduce_all()
-        self.inputs = {'X': np.random.random((5, 6, 7, 8)).astype("float32")}
+        self.inputs = {'X': np.random.random((5, 6, 7, 8)).astype('float32')}
         self.attrs = {
             'dim': 2,
             'keep_dim': self.keep_dim,
@@ -31,7 +31,7 @@ def setUp(self):
         self.outputs = {'Out': np.zeros((1, 1))}
 
     def init_op_type(self):
-        self.op_type = "reduce_sum"
+        self.op_type = 'reduce_sum'
 
     def init_keep_dim(self):
         self.keep_dim = True
@@ -45,7 +45,7 @@ def test_check_output(self):
 
 class TestReduceMeanOp(TestReduceSumOp):
     def init_op_type(self):
-        self.op_type = "reduce_mean"
+        self.op_type = 'reduce_mean'
 
     def init_reduce_all(self):
         self.reduce_all = True
@@ -53,7 +53,7 @@ def init_reduce_all(self):
 
 class TestReduceMaxOp(TestReduceSumOp):
     def init_op_type(self):
-        self.op_type = "reduce_max"
+        self.op_type = 'reduce_max'
 
     def init_keep_dim(self):
         self.keep_dim = False
@@ -64,7 +64,7 @@ def init_reduce_all(self):
 
 class TestReduceMinOp(TestReduceSumOp):
     def init_op_type(self):
-        self.op_type = "reduce_min"
+        self.op_type = 'reduce_min'
 
     def init_keep_dim(self):
         self.keep_dim = False