diff --git a/python/paddle/fluid/contrib/layers/nn.py b/python/paddle/fluid/contrib/layers/nn.py
index 7ab225c8c9f825d60c7e910be92f0190106f3860..b836dfa451c33b57c873f5fcc019b40985d005d4 100644
--- a/python/paddle/fluid/contrib/layers/nn.py
+++ b/python/paddle/fluid/contrib/layers/nn.py
@@ -98,10 +98,12 @@ def fused_embedding_seq_pool(
         .. code-block:: python
             import numpy as np
             import paddle.fluid as fluid
+            import paddle
+            paddle.enable_static()
 
             dict_size = 20
-            data_t = fluid.layers.data(
-                name='word', shape=[1], dtype='int64', lod_level=1)
+            data_t = paddle.static.data(
+                name='word', shape=[-1, 1], dtype='int64', lod_level=1)
             padding_idx = np.random.randint(1, 10)
             out = fluid.contrib.fused_embedding_seq_pool(
                 input=data_t,
@@ -305,11 +307,13 @@ def multiclass_nms2(
 
 
             import paddle.fluid as fluid
-            boxes = fluid.layers.data(name='bboxes', shape=[81, 4],
+            import paddle
+            paddle.enable_static()
+            boxes = paddle.static.data(name='bboxes', shape=[-1, 81, 4],
                                       dtype='float32', lod_level=1)
-            scores = fluid.layers.data(name='scores', shape=[81],
+            scores = paddle.static.data(name='scores', shape=[-1, 81],
                                       dtype='float32', lod_level=1)
-            out, index = fluid.layers.multiclass_nms2(bboxes=boxes,
+            out, index = fluid.contrib.layers.multiclass_nms2(bboxes=boxes,
                                               scores=scores,
                                               background_label=0,
                                               score_threshold=0.5,
@@ -501,7 +505,9 @@ def shuffle_batch(x, seed=None):
         .. code-block:: python
 
             import paddle.fluid as fluid
-            x = fluid.layers.data(name="x", shape=[-1, 4])
+            import paddle
+            paddle.enable_static()
+            x = paddle.static.data(name="x", shape=[-1, 4])
             out = fluid.contrib.layers.shuffle_batch(x)
     """
     helper = LayerHelper('shuffle_batch', **locals())
@@ -1313,7 +1319,7 @@ def _pull_box_extended_sparse(input, size, extend_size=64, dtype='float32'):
     Examples:
         .. code-block:: python
           import paddle.fluid as fluid
-          data = fluid.layers.data(name='sequence', shape=[1], dtype='int64', lod_level=1)
+          data = paddle.static.data(name='sequence', shape=[-1, 1], dtype='int64', lod_level=1)
           emb, emb_ex = fluid.contrib.layers._pull_box_extended_sparse(input=data, size=8, extend_size=128)
     """
     helper = LayerHelper('pull_box_extended_sparse', **locals())
@@ -1438,15 +1444,14 @@ def correlation(
         .. code-block:: python
 
             import paddle.fluid as fluid
-
-            x1 = fluid.layers.data(name='x1',
-                               shape=x_shape,
-                               dtype=x_type,
-                               append_batch_size=False)
-            x2 = fluid.layers.data(name='x2',
-                                shape=x_shape,
-                                dtype=x_type,
-                                append_batch_size=False)
+            import paddle
+            paddle.enable_static()
+            x1 = paddle.static.data(name='x1',
+                               shape=[2,3,4,5],
+                               dtype="float32")
+            x2 = paddle.static.data(name='x2',
+                                shape=[2,3,4,5],
+                                dtype="float32")
 
 
             out = fluid.contrib.correlation(
@@ -1555,8 +1560,8 @@ def fused_bn_add_act(
             # required: gpu
             def build_program(main_program, startup_program):
                 with fluid.program_guard(main_program, startup_program):
-                    x = fluid.layers.data(name='x', shape=[1, 28, 28], dtype='float32')
-                    y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+                    x = paddle.static.data(name='x', shape=[-1, 1, 28, 28], dtype='float32')
+                    y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
                     conv1_1 = paddle.static.nn.conv2d(
                         input=x,
                         filter_size=3,
diff --git a/python/paddle/fluid/contrib/tests/test_correlation.py b/python/paddle/fluid/contrib/tests/test_correlation.py
index 4e9ef9b0fe8f554e71ce98c8d7b33661abc2fd6c..0f614f4324819a8bbaba8f7b6a85aae7d04e0249 100644
--- a/python/paddle/fluid/contrib/tests/test_correlation.py
+++ b/python/paddle/fluid/contrib/tests/test_correlation.py
@@ -85,20 +85,20 @@ class TestCorrelationOp(unittest.TestCase):
         np.set_printoptions(threshold=np.inf)
         x_shape = (2, 10, 3, 3)
         x_type = 'float32'
-        x1 = fluid.layers.data(
+        x1 = paddle.static.data(
             name='x1',
             shape=x_shape,
             dtype=x_type,
-            append_batch_size=False,
-            stop_gradient=False,
         )
-        x2 = fluid.layers.data(
+        x1.desc.set_need_check_feed(False)
+        x1.stop_gradient = False
+        x2 = paddle.static.data(
             name='x2',
             shape=x_shape,
             dtype=x_type,
-            append_batch_size=False,
-            stop_gradient=False,
         )
+        x2.desc.set_need_check_feed(False)
+        x2.stop_gradient = False
 
         x1_np = np.random.randn(2, 3, 4, 5).astype(x_type)
         x2_np = np.random.randn(2, 3, 4, 5).astype(x_type)
diff --git a/python/paddle/fluid/contrib/tests/test_image_classification_fp16.py b/python/paddle/fluid/contrib/tests/test_image_classification_fp16.py
index f99d640799764af631d02a2faef170a46de971e6..5ab54bc7161780859445e31bd3086654a9ff09c5 100644
--- a/python/paddle/fluid/contrib/tests/test_image_classification_fp16.py
+++ b/python/paddle/fluid/contrib/tests/test_image_classification_fp16.py
@@ -110,10 +110,10 @@ def train(net_type, use_cuda, save_dirname, is_local):
     train_program.random_seed = 123
     startup_prog.random_seed = 456
     with fluid.program_guard(train_program, startup_prog):
-        images = fluid.layers.data(
-            name='pixel', shape=data_shape, dtype='float32'
+        images = paddle.static.data(
+            name='pixel', shape=[-1] + data_shape, dtype='float32'
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         if net_type == "vgg":
             print("train vgg net")
@@ -444,11 +444,11 @@ class TestAmpWithNonIterableDataLoader(unittest.TestCase):
         start_prog = paddle.static.Program()
         with paddle.static.program_guard(main_prog, start_prog):
             with paddle.fluid.unique_name.guard():
-                image = fluid.layers.data(
-                    name='image', shape=[3, 224, 224], dtype='float32'
+                image = paddle.static.data(
+                    name='image', shape=[-1, 3, 224, 224], dtype='float32'
                 )
-                label = fluid.layers.data(
-                    name='label', shape=[1], dtype='int64'
+                label = paddle.static.data(
+                    name='label', shape=[-1, 1], dtype='int64'
                 )
                 py_reader = fluid.io.DataLoader.from_generator(
                     feed_list=[image, label],
diff --git a/python/paddle/fluid/contrib/tests/test_model_cast_to_bf16.py b/python/paddle/fluid/contrib/tests/test_model_cast_to_bf16.py
index 7254dd9df31821c2c8653676f22c98030943aa2b..a97638536891873245ab249ce16f7b89ab4eebd5 100644
--- a/python/paddle/fluid/contrib/tests/test_model_cast_to_bf16.py
+++ b/python/paddle/fluid/contrib/tests/test_model_cast_to_bf16.py
@@ -96,14 +96,22 @@ class TestModelCastBF16(unittest.TestCase):
         nn_bf16 = amp.bf16.convert_float_to_uint16(nn)
 
         with self.static_graph():
-            t_bf16 = layers.data(
-                name='t_bf16', shape=[size, size], dtype=np.int32
+            t_bf16 = paddle.static.data(
+                name='t_bf16', shape=[-1, size, size], dtype='int32'
             )
-            tt_bf16 = layers.data(
-                name='tt_bf16', shape=[size, size], dtype=np.int32
+            t_bf16.desc.set_need_check_feed(False)
+            tt_bf16 = paddle.static.data(
+                name='tt_bf16', shape=[-1, size, size], dtype='int32'
             )
-            t = layers.data(name='t', shape=[size, size], dtype='float32')
-            tt = layers.data(name='tt', shape=[size, size], dtype='float32')
+            tt_bf16.desc.set_need_check_feed(False)
+            t = paddle.static.data(
+                name='t', shape=[-1, size, size], dtype='float32'
+            )
+            t.desc.set_need_check_feed(False)
+            tt = paddle.static.data(
+                name='tt', shape=[-1, size, size], dtype='float32'
+            )
+            tt.desc.set_need_check_feed(False)
 
             ret = paddle.add(t, tt)
             ret = paddle.multiply(ret, t)
@@ -143,8 +151,14 @@ class TestModelCastBF16(unittest.TestCase):
         )
 
         with self.static_graph():
-            t = layers.data(name='t', shape=[size, size], dtype='float32')
-            tt = layers.data(name='tt', shape=[size, size], dtype='float32')
+            t = paddle.static.data(
+                name='t', shape=[-1, size, size], dtype='float32'
+            )
+            t.desc.set_need_check_feed(False)
+            tt = paddle.static.data(
+                name='tt', shape=[-1, size, size], dtype='float32'
+            )
+            tt.desc.set_need_check_feed(False)
 
             with amp.bf16.bf16_guard():
                 ret = paddle.add(t, tt)
diff --git a/python/paddle/fluid/contrib/tests/test_multi_precision_fp16_train.py b/python/paddle/fluid/contrib/tests/test_multi_precision_fp16_train.py
index 1d4873817ab3a2b1613dea3ec578fc0cf48e9ab5..75754fb8bb6a88706393e20578e369958608e346 100644
--- a/python/paddle/fluid/contrib/tests/test_multi_precision_fp16_train.py
+++ b/python/paddle/fluid/contrib/tests/test_multi_precision_fp16_train.py
@@ -102,10 +102,10 @@ def train(use_pure_fp16=True, use_nesterov=False, optimizer=""):
     train_program.random_seed = 123
     startup_prog.random_seed = 456
     with fluid.program_guard(train_program, startup_prog):
-        images = fluid.layers.data(
-            name='pixel', shape=data_shape, dtype='float32'
+        images = paddle.static.data(
+            name='pixel', shape=[-1] + data_shape, dtype='float32'
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         net = resnet_cifar10(images)
         logits = paddle.static.nn.fc(x=net, size=classdim, activation="softmax")
         cost = paddle.nn.functional.softmax_with_cross_entropy(
@@ -275,11 +275,11 @@ class TestAmpWithNonIterableDataLoader(unittest.TestCase):
         start_prog = paddle.static.Program()
         with paddle.static.program_guard(main_prog, start_prog):
             with paddle.fluid.unique_name.guard():
-                image = fluid.layers.data(
-                    name='image', shape=[3, 224, 224], dtype='float32'
+                image = paddle.static.data(
+                    name='image', shape=[-1, 3, 224, 224], dtype='float32'
                 )
-                label = fluid.layers.data(
-                    name='label', shape=[1], dtype='int64'
+                label = paddle.static.data(
+                    name='label', shape=[-1, 1], dtype='int64'
                 )
                 py_reader = fluid.io.DataLoader.from_generator(
                     feed_list=[image, label],
diff --git a/python/paddle/fluid/contrib/tests/test_weight_decay_extend.py b/python/paddle/fluid/contrib/tests/test_weight_decay_extend.py
index 7af54b7d1573830ebdb244ba520bccf76ad67638..870aceb5571aa892e160d56ebcea88f4a9cccaed 100644
--- a/python/paddle/fluid/contrib/tests/test_weight_decay_extend.py
+++ b/python/paddle/fluid/contrib/tests/test_weight_decay_extend.py
@@ -134,10 +134,12 @@ class TestWeightDecay(unittest.TestCase):
         startup_prog = fluid.framework.Program()
 
         with prog_scope_guard(main_prog=main_prog, startup_prog=startup_prog):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
             avg_cost = model(data, label, self.word_dict_len)
             AdamW = fluid.contrib.extend_with_decoupled_weight_decay(
                 fluid.optimizer.Adam
@@ -158,10 +160,12 @@ class TestWeightDecay(unittest.TestCase):
         startup_prog = fluid.framework.Program()
 
         with prog_scope_guard(main_prog=main_prog, startup_prog=startup_prog):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
 
             avg_cost = model(data, label, self.word_dict_len)
 
diff --git a/python/paddle/fluid/executor.py b/python/paddle/fluid/executor.py
index e3376d8446586607947bf2143fe5c9fe32115dac..2822a87a02172e22c2dc7c6899b7db4f5a7eea0f 100755
--- a/python/paddle/fluid/executor.py
+++ b/python/paddle/fluid/executor.py
@@ -1862,7 +1862,6 @@ class Executor:
                 vardesc = global_block.desc.find_var(varname.encode())
                 varobj = global_block.vars[varname]
 
-                # Can not check var build by fluid.layers.data(), bucause fluid.layers.data() had not set need_check_feed
                 if (
                     vardesc.persistable() == False
                     and vardesc.type() == core.VarDesc.VarType.LOD_TENSOR
diff --git a/python/paddle/fluid/framework.py b/python/paddle/fluid/framework.py
index da2fa96c758bae003952b6c2181b35974231c0bd..8b019114a322b9fd19e18f499e60d574002c7dd9 100644
--- a/python/paddle/fluid/framework.py
+++ b/python/paddle/fluid/framework.py
@@ -6957,9 +6957,10 @@ class Parameter(Variable, metaclass=ParameterMetaClass):
             .. code-block:: python
 
                 import paddle.fluid as fluid
+                import paddle
 
                 prog = fluid.default_main_program()
-                rlt = fluid.layers.data("fake_data", shape=[1,1], dtype='float32')
+                rlt = paddle.static.data("fake_data", shape=[-1,1,1], dtype='float32')
                 debug_str = prog.to_string(throw_on_error=True, with_details=False)
                 print(debug_str)
         """
diff --git a/python/paddle/fluid/incubate/fleet/tests/fleet_deep_ctr.py b/python/paddle/fluid/incubate/fleet/tests/fleet_deep_ctr.py
index e2b79936481b643fb9a3e04a0871de5f519346b0..23f5a44fe139e743d13352aea00029c355c50853 100644
--- a/python/paddle/fluid/incubate/fleet/tests/fleet_deep_ctr.py
+++ b/python/paddle/fluid/incubate/fleet/tests/fleet_deep_ctr.py
@@ -80,26 +80,23 @@ def model():
         train_file_path,
     ) = ctr_dataset_reader.prepare_data()
     """ network definition """
-    dnn_data = fluid.layers.data(
+    dnn_data = paddle.static.data(
         name="dnn_data",
         shape=[-1, 1],
         dtype="int64",
         lod_level=1,
-        append_batch_size=False,
     )
-    lr_data = fluid.layers.data(
+    lr_data = paddle.static.data(
         name="lr_data",
         shape=[-1, 1],
         dtype="int64",
         lod_level=1,
-        append_batch_size=False,
     )
-    label = fluid.layers.data(
+    label = paddle.static.data(
         name="click",
         shape=[-1, 1],
         dtype="int64",
         lod_level=0,
-        append_batch_size=False,
     )
 
     datas = [dnn_data, lr_data, label]
diff --git a/python/paddle/fluid/incubate/fleet/utils/fleet_util.py b/python/paddle/fluid/incubate/fleet/utils/fleet_util.py
index 4ec3c1d16e077ea00672c664bac3b1b4ea5e491c..e0ae707be934d9dad09faf87914ce5714fb0f20a 100644
--- a/python/paddle/fluid/incubate/fleet/utils/fleet_util.py
+++ b/python/paddle/fluid/incubate/fleet/utils/fleet_util.py
@@ -1371,8 +1371,8 @@ class FleetUtil:
                                                           local_total_ins.name)
 
               # below is part of example model
-              label = fluid.layers.data(name="click", shape=[-1, 1],\
-                  dtype="int64", lod_level=0, append_batch_size=False)
+              label = paddle.static.data(name="click", shape=[-1, 1],\
+                  dtype="int64", lod_level=0)
               emb = my_slot_net(slots, label) # emb can be fc layer of size 1
               similarity_norm = fluid.layers.sigmoid(paddle.clip(\
                   emb, min=-15.0, max=15.0), name="similarity_norm")\
@@ -1571,8 +1571,8 @@ class FleetUtil:
                                               local_total_ins.name)
 
               # below is part of model
-              label = fluid.layers.data(name="click", shape=[-1, 1],\
-                  dtype="int64", lod_level=0, append_batch_size=False)
+              label = paddle.static.data(name="click", shape=[-1, 1],\
+                  dtype="int64", lod_level=0)
               emb = my_slot_net(slots, label) # emb can be fc layer of size 1
               similarity_norm = fluid.layers.sigmoid(paddle.clip(\
                   emb, min=-15.0, max=15.0), name="similarity_norm")\
diff --git a/python/paddle/fluid/incubate/fleet/utils/utils.py b/python/paddle/fluid/incubate/fleet/utils/utils.py
index ef022c96ecf8650a8bf2b4816ef666ec42ebd006..4aff834b12eaed276bbfb6a15ec1fa3b81b02cd6 100644
--- a/python/paddle/fluid/incubate/fleet/utils/utils.py
+++ b/python/paddle/fluid/incubate/fleet/utils/utils.py
@@ -17,6 +17,7 @@ import sys
 import logging
 import subprocess
 import numpy as np
+import paddle
 from collections import OrderedDict
 import paddle.fluid as fluid
 from paddle.fluid import core
@@ -172,8 +173,9 @@ def save_var(np_array, var_name, shape_list, dtype, save_path):
     program = fluid.Program()
     place = fluid.CPUPlace()
     exe = fluid.Executor(place)
+    shape = list(shape_list)
     with fluid.program_guard(program):
-        d0_data = fluid.layers.data(var_name, shape=shape_list, dtype=dtype)
+        d0_data = paddle.static.data(var_name, shape=shape, dtype=dtype)
         append_save_op(program.global_block(), d0_data, save_path)
         exe.run(feed={var_name: np_array}, fetch_list=[])
 
@@ -183,7 +185,7 @@ def load_var(var_name, shape_list, dtype, save_path):
     place = fluid.CPUPlace()
     exe = fluid.Executor(place)
     with fluid.program_guard(program):
-        d0_data = fluid.layers.data(var_name, shape=shape_list, dtype=dtype)
+        d0_data = paddle.static.data(var_name, shape=shape_list, dtype=dtype)
         append_load_op(program.global_block(), d0_data, save_path)
         outs = exe.run(feed={}, fetch_list=[d0_data])
         return outs
diff --git a/python/paddle/fluid/install_check.py b/python/paddle/fluid/install_check.py
index bf1ad9b107f74694c80472f583287d617fdf0616..241dd71e200aba975b4660afbcdf078d61fad2dc 100644
--- a/python/paddle/fluid/install_check.py
+++ b/python/paddle/fluid/install_check.py
@@ -103,7 +103,7 @@ def run_check():
                 with unique_name.guard():
                     build_strategy = compiler.BuildStrategy()
                     build_strategy.enable_inplace = True
-                    inp = layers.data(name="inp", shape=[2, 2])
+                    inp = paddle.static.data(name="inp", shape=[-1, 2, 2])
                     simple_layer = SimpleLayer(input_size=2)
                     out = simple_layer(inp)
                     exe = executor.Executor(
@@ -138,9 +138,7 @@ def run_check():
         with executor.scope_guard(scope):
             with program_guard(train_prog, startup_prog):
                 with unique_name.guard():
-                    inp0 = layers.data(
-                        name="inp", shape=[2, 2], append_batch_size=False
-                    )
+                    inp0 = paddle.static.data(name="inp", shape=[2, 2])
                     simple_layer0 = SimpleLayer(input_size=2)
                     out0 = simple_layer0(inp0)
                     param_grads = backward.append_backward(
diff --git a/python/paddle/fluid/io.py b/python/paddle/fluid/io.py
index 490e9412cb2a76ca8c1535928a26f9227071bd37..9faeacee886c46c277d0e3240eb5263af822dd54 100644
--- a/python/paddle/fluid/io.py
+++ b/python/paddle/fluid/io.py
@@ -355,7 +355,7 @@ def save_vars(
             main_prog = fluid.Program()
             startup_prog = fluid.Program()
             with fluid.program_guard(main_prog, startup_prog):
-                data = fluid.layers.data(name="img", shape=[64, 784], append_batch_size=False)
+                data = paddle.static.data(name="img", shape=[64, 784])
                 w = paddle.create_parameter(shape=[784, 200], dtype='float32', name='fc_w')
                 b = paddle.create_parameter(shape=[200], dtype='float32', name='fc_b')
                 hidden_w = paddle.matmul(x=data, y=w)
@@ -830,7 +830,7 @@ def load_vars(
             main_prog = fluid.Program()
             startup_prog = fluid.Program()
             with fluid.program_guard(main_prog, startup_prog):
-                data = fluid.layers.data(name="img", shape=[64, 784], append_batch_size=False)
+                data = paddle.static.data(name="img", shape=[64, 784])
                 w = paddle.create_parameter(shape=[784, 200], dtype='float32', name='fc_w')
                 b = paddle.create_parameter(shape=[200], dtype='float32', name='fc_b')
                 hidden_w = paddle.matmul(x=data, y=w)
@@ -1598,7 +1598,7 @@ def load_inference_model(
             main_prog = fluid.Program()
             startup_prog = fluid.Program()
             with fluid.program_guard(main_prog, startup_prog):
-                data = fluid.layers.data(name="img", shape=[64, 784], append_batch_size=False)
+                data = paddle.static.data(name="img", shape=[-1, 64, 784])
                 w = paddle.create_parameter(shape=[784, 200], dtype='float32')
                 b = paddle.create_parameter(shape=[200], dtype='float32')
                 hidden_w = paddle.matmul(x=data, y=w)
diff --git a/python/paddle/fluid/layers/control_flow.py b/python/paddle/fluid/layers/control_flow.py
index e01536b2e38293147ff9cb163d29e32300c14d22..5c877c1e8f176ce50e7f54e308a49e8a89912bcd 100755
--- a/python/paddle/fluid/layers/control_flow.py
+++ b/python/paddle/fluid/layers/control_flow.py
@@ -466,7 +466,7 @@ class StaticRNN:
                         is_sparse=False)
                 # transform batch size to dim 1
                 x_emb = paddle.transpose(x_emb, perm=[1, 0, 2])
-                boot_memory = fluid.layers.data(name='boot', shape=[hidden_size], dtype='float32', lod_level=1)
+                boot_memory = paddle.static.data(name='boot', shape=[-1, hidden_size], dtype='float32', lod_level=1)
                 rnn = fluid.layers.StaticRNN()
                 with rnn.step():
                         # mark created x_emb as input, each step process a word
diff --git a/python/paddle/fluid/layers/io.py b/python/paddle/fluid/layers/io.py
index e33cf23e6c1bc15f490b22f80ab0b32efe70d701..994fc98038086cbe581ed8e30ea183a75cc168f5 100644
--- a/python/paddle/fluid/layers/io.py
+++ b/python/paddle/fluid/layers/io.py
@@ -41,108 +41,7 @@ from ..framework import (
     _set_expected_place,
 )
 
-__all__ = [
-    'data',
-]
-
-
-@static_only
-def data(
-    name,
-    shape,
-    append_batch_size=True,
-    dtype='float32',
-    lod_level=0,
-    type=core.VarDesc.VarType.LOD_TENSOR,
-    stop_gradient=True,
-):
-    """
-    **Data Layer**
-
-    This operator creates the global variable. The global variables can be
-    accessed by all the following operators in the graph.
-
-    Note:
-        :code:`paddle.fluid.layers.data` is deprecated as it will be removed in
-        a later version. Please use :code:`paddle.fluid.data` .
-
-        This :code:`paddle.fluid.layers.data` set shape and dtype at compile
-        time but does NOT check the shape or the dtype of fed data, the
-        :code:`paddle.fluid.data` checks the shape and the dtype of data fed
-        by Executor or ParallelExecutor during run time.
-
-        To feed variable size inputs, users can feed variable size inputs
-        directly to this :code:`paddle.fluid.layers.data` and PaddlePaddle will
-        fit the size accordingly. Or set -1 on the variable dimension when using
-        :code:`paddle.fluid.data` .
-
-        The default :code:`stop_gradient` attribute of the Variable created by
-        this API is true, which means the gradient won't be passed backward
-        through the data Varaible. Set :code:`var.stop_gradient = False` If
-        user would like to pass backward gradient.
-
-    Args:
-       name(str): The name/alias of the variable, see :ref:`api_guide_Name`
-            for more details.
-       shape(list|tuple): Tuple declaring the shape. If :code:`append_batch_size` is
-            True and there is no -1 inside :code:`shape`, it should be
-            considered as the shape of the each sample. Otherwise, it should
-            be considered as the shape of the batched data.
-       append_batch_size(bool):
-          1. If true, it prepends -1 to the shape.
-            For example if shape=[1], the resulting shape is [-1, 1]. This will
-            be useful to set different batch size at run time.
-          2. If shape contains -1, such as shape=[1, -1].
-            append_batch_size will be enforced to be be False (ineffective)
-            because PaddlePaddle cannot set more than 1 unknown number on the
-            shape.
-       dtype(np.dtype|VarType|str): The type of the data. Supported dtype: bool,
-            float16, float32, float64, int8, int16, int32, int64, uint8.
-       type(VarType): The output type. Supported dtype: VarType.LOD_TENSOR,
-            VarType.SELECTED_ROWS, VarType.NCCL_ID. Default: VarType.LOD_TENSOR.
-       lod_level(int): The LoD Level. 0 means the input data is not a sequence.
-            Default: 0.
-       stop_gradient(bool): A boolean that mentions whether gradient should flow.
-            Default: True.
-
-    Returns:
-        The global variable that gives access to the data.
-
-    Return Type:
-        Variable
-
-    Examples:
-        .. code-block:: python
-
-          import paddle.fluid as fluid
-          data = fluid.layers.data(name='x', shape=[784], dtype='float32')
-    """
-    helper = LayerHelper('data', **locals())
-
-    check_type(name, 'name', (bytes, str), 'data')
-    check_type(shape, 'shape', (list, tuple), 'data')
-
-    shape = list(shape)
-    for i in range(len(shape)):
-        if shape[i] is None:
-            shape[i] = -1
-            append_batch_size = False
-        elif shape[i] < 0:
-            append_batch_size = False
-
-    if append_batch_size:
-        shape = [-1] + shape  # append batch size as -1
-
-    data_var = helper.create_global_variable(
-        name=name,
-        shape=shape,
-        dtype=dtype,
-        type=type,
-        stop_gradient=stop_gradient,
-        lod_level=lod_level,
-        is_data=True,
-    )
-    return data_var
+__all__ = []
 
 
 class BlockGuardServ(BlockGuard):
@@ -189,11 +88,10 @@ class ListenAndServ:
                 serv = layers.ListenAndServ(
                     "127.0.0.1:6170", ["X"], optimizer_mode=False)
                 with serv.do():
-                    x = layers.data(
+                    x = paddle.static.data(
                         shape=[32, 32],
                         dtype='float32',
-                        name="X",
-                        append_batch_size=False)
+                        name="X")
                     fluid.initializer.Constant(value=1.0)(x, main.global_block())
                     paddle.scale(x=x, scale=10.0, out=out_var)
 
diff --git a/python/paddle/fluid/layers/nn.py b/python/paddle/fluid/layers/nn.py
index fa0e4007eaa76ef172522dcda4aec1ad93948a26..fa0f49d01b99796c47f33269775ad73cc4c186eb 100644
--- a/python/paddle/fluid/layers/nn.py
+++ b/python/paddle/fluid/layers/nn.py
@@ -329,7 +329,7 @@ def _pull_sparse(
         .. code-block:: python
 
           import paddle.fluid as fluid
-          data = fluid.layers.data(name='sequence', shape=[1], dtype='int64', lod_level=1)
+          data = paddle.static.data(name='sequence', shape=[-1, 1], dtype='int64', lod_level=1)
           emb = fluid.layers.nn._pull_sparse(
               input=data, size=11, table_id=0, accessor_class="DownpourCtrAccessor")
     """
@@ -403,7 +403,7 @@ def _pull_sparse_v2(
         .. code-block:: python
 
           import paddle.fluid as fluid
-          data = fluid.layers.data(name='sequence', shape=[1], dtype='int64', lod_level=1)
+          data = paddle.static.data(name='sequence', shape=[-1, 1], dtype='int64', lod_level=1)
           emb = fluid.layers.nn._pull_sparse_v2(
               input=data, size=11, table_id=0, accessor_class="DownpourCtrAccessor")
     """
@@ -464,9 +464,9 @@ def _pull_gpups_sparse(
 
           import paddle.fluid as fluid
           slots = []
-          data_1 = fluid.layers.data(name='sequence', shape=[1], dtype='int64', lod_level=1)
+          data_1 = paddle.static.data(name='sequence', shape=[-1,1], dtype='int64', lod_level=1)
           slots.append(data_1)
-          data_2 = fluid.layers.data(name='sequence', shape=[1], dtype='int64', lod_level=1)
+          data_2 = paddle.static.data(name='sequence', shape=[-1,1], dtype='int64', lod_level=1)
           slots.append(data_2)
           embs = fluid.layers.pull_gpups_sparse(input=slots, size=[11, 35])
     """
@@ -526,7 +526,7 @@ def _pull_box_sparse(
         .. code-block:: python
 
           import paddle.fluid as fluid
-          data = fluid.layers.data(name='sequence', shape=[1], dtype='int64', lod_level=1)
+          data = paddle.static.data(name='sequence', shape=[-1,1], dtype='int64', lod_level=1)
           emb = fluid.layers.pull_box_sparse(input=data, size=[11])
     """
     helper = LayerHelper('pull_box_sparse', **locals())
@@ -711,7 +711,7 @@ def unsqueeze(input, axes, name=None):
         .. code-block:: python
 
             import paddle.fluid as fluid
-            x = fluid.layers.data(name='x', shape=[5, 10])
+            x = paddle.static.data(name='x', shape=[-1, 5, 10], dtype="float32")
             y = fluid.layers.unsqueeze(input=x, axes=[1])
 
     """
diff --git a/python/paddle/fluid/optimizer.py b/python/paddle/fluid/optimizer.py
index 1bab4f34317719ae4d806a5ca73bd4ac7ae86ed8..52501992ef9e98f435d19af941e8e9351ebae810 100755
--- a/python/paddle/fluid/optimizer.py
+++ b/python/paddle/fluid/optimizer.py
@@ -1431,8 +1431,8 @@ class SGDOptimizer(Optimizer):
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
-                x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-                y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+                x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+                y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
                 y_predict = paddle.static.nn.fc(x, size=1, activation=None)
                 cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = paddle.mean(cost)
@@ -1623,8 +1623,8 @@ class MomentumOptimizer(Optimizer):
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
-                x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-                y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+                x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+                y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
                 y_predict = paddle.static.nn.fc(x, size=1, activation=None)
                 cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = paddle.mean(cost)
@@ -1772,8 +1772,8 @@ class LarsMomentumOptimizer(Optimizer):
 
             paddle.enable_static()
             np_inp = np.array([[1.0, 2.0], [3.0, 4.0]], dtype=np.float32)
-            inp = fluid.layers.data(
-                name="inp", shape=[2, 2], append_batch_size=False)
+            inp = paddle.static.data(
+                name="inp", shape=[2, 2], dtype='float32')
             out = paddle.static.nn.fc(inp, size=3)
             out = paddle.sum(out)
             optimizer = fluid.optimizer.LarsMomentumOptimizer(learning_rate=0.001, momentum=0.9)
@@ -2764,7 +2764,7 @@ class DpsgdOptimizer(Optimizer):
           train_program = fluid.Program()
           startup_program = fluid.Program()
           with fluid.program_guard(train_program, startup_program):
-              data = fluid.layers.data(name='X', shape=[1], dtype='float32')
+              data = paddle.static.data(name='X', shape=[-1,1], dtype='float32')
               hidden = paddle.static.nn.fc(x=data, size=10)
               loss = paddle.mean(hidden)
               optimizer = fluid.optimizer.Dpsgd(learning_rate=0.01, clip=10.0, batch_size=16.0, sigma=1.0)
@@ -3217,8 +3217,8 @@ class RMSPropOptimizer(Optimizer):
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
-                x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-                y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+                x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+                y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
                 y_predict = paddle.static.nn.fc(x, size=1, activation=None)
                 cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = paddle.mean(cost)
@@ -3415,8 +3415,8 @@ class FtrlOptimizer(Optimizer):
             place = fluid.CPUPlace()
             main = fluid.Program()
             with fluid.program_guard(main):
-                x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-                y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+                x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+                y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
                 y_predict = paddle.static.nn.fc(x, size=1, activation=None)
                 cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
                 avg_cost = paddle.mean(cost)
@@ -4354,11 +4354,12 @@ class PipelineOptimizer:
             import paddle
             import paddle.fluid as fluid
             import paddle.fluid.layers as layers
+            import numpy as np
 
             paddle.enable_static()
             with fluid.device_guard("gpu:0"):
-                x = fluid.layers.data(name='x', shape=[1], dtype='int64', lod_level=0)
-                y = fluid.layers.data(name='y', shape=[1], dtype='int64', lod_level=0)
+                x = paddle.static.data(name='x', shape=[-1, 1], dtype='int64', lod_level=0)
+                y = paddle.static.data(name='y', shape=[-1, 1], dtype='int64', lod_level=0)
                 data_loader = fluid.io.DataLoader.from_generator(
                     feed_list=[x, y],
                     capacity=64,
@@ -6332,8 +6333,8 @@ class RecomputeOptimizer(Optimizer):
                 )
                 sum_cost = paddle.mean(cost)
                 return sum_cost, fc_1, prediction
-            input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-            input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+            input_x = paddle.static.data(name="x", shape=[-1,32], dtype='float32')
+            input_y = paddle.static.data(name="y", shape=[-1,1], dtype='int64')
             cost, fc_1, pred = mlp(input_x, input_y)
 
             sgd = fluid.optimizer.Adam(learning_rate=0.01)
@@ -6410,8 +6411,8 @@ class RecomputeOptimizer(Optimizer):
                     sum_cost = paddle.mean(cost)
                     return sum_cost, fc_1, prediction
 
-                input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-                input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+                input_x = paddle.static.data(name="x", shape=[-1,32], dtype='float32')
+                input_y = paddle.static.data(name="y", shape=[-1,1], dtype='int64')
                 cost, fc_1, pred = mlp(input_x, input_y)
                 print("Finished FF")
 
@@ -6458,8 +6459,8 @@ class RecomputeOptimizer(Optimizer):
                     return sum_cost, fc_1, prediction
 
 
-                input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-                input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+                input_x = paddle.static.data(name="x", shape=[-1,32], dtype='float32')
+                input_y = paddle.static.data(name="y", shape=[-1,1], dtype='int64')
                 cost, fc_1, pred = mlp(input_x, input_y)
                 print("Finished FF")
 
@@ -6952,8 +6953,8 @@ class RecomputeOptimizer(Optimizer):
                     return sum_cost, fc_1, prediction
 
 
-                input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-                input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+                input_x = paddle.static.data(name="x", shape=[-1,32], dtype='float32')
+                input_y = paddle.static.data(name="y", shape=[-1,1], dtype='int64')
                 cost, fc_1, pred = mlp(input_x, input_y)
                 print("Finished FF")
 
@@ -7033,8 +7034,8 @@ class RecomputeOptimizer(Optimizer):
                     sum_cost = paddle.mean(cost)
                     return sum_cost, fc_1, prediction
 
-                input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-                input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+                input_x = paddle.static.data(name="x", shape=[-1,32], dtype='float32')
+                input_y = paddle.static.data(name="y", shape=[-1,1], dtype='int64')
                 cost, fc_1, pred = mlp(input_x, input_y)
                 print("Finished FF")
 
@@ -7120,8 +7121,8 @@ class LookaheadOptimizer:
 
             paddle.enable_static()
 
-            x = fluid.layers.data(name='x', shape=[2], dtype='float32')
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+            x = paddle.static.data(name='x', shape=[-1,2], dtype='float32')
+            label = paddle.static.data(name="label", shape=[-1,1], dtype="int64")
             y = paddle.static.nn.fc(x=[x], size=2, activation="softmax")
             loss = paddle.nn.functional.cross_entropy(
                 input=y, label=label,
@@ -7311,8 +7312,8 @@ class GradientMergeOptimizer:
             sum_cost = paddle.mean(cost)
             return sum_cost, fc_1, prediction
 
-        input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-        input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1,32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1,1], dtype='int64')
         cost, fc_1, pred = mlp(input_x, input_y)
         sgd = fluid.optimizer.Adam(learning_rate=0.01)
         sgd = fluid.optimizer.GradientMergeOptimizer(sgd, k_steps=4, avg=True)
diff --git a/python/paddle/fluid/reader.py b/python/paddle/fluid/reader.py
index e6cabbdde92683eaaa19210bbad5f373bf420b24..8e36dd28005be7cf0c21423ae93d8e6a62c2d88e 100644
--- a/python/paddle/fluid/reader.py
+++ b/python/paddle/fluid/reader.py
@@ -1643,7 +1643,7 @@ class PyReader(DataLoaderBase):
 
     Args:
         feed_list (list(Variable)|tuple(Variable)): feed variable list.
-            The variables should be created by :code:`fluid.layers.data()`.
+            The variables should be created by :code:`paddle.static.data()`.
         capacity (int): capacity of the queue maintained in PyReader.
             The unit is batch number. Set larger capacity if your reader
             is fast.
diff --git a/python/paddle/fluid/regularizer.py b/python/paddle/fluid/regularizer.py
index 36e648e3e38edc6d733f7d9952519a6189db104b..84bfa351c962db527c72d68dd77ad89b4e855336 100644
--- a/python/paddle/fluid/regularizer.py
+++ b/python/paddle/fluid/regularizer.py
@@ -74,8 +74,8 @@ class L2DecayRegularizer(WeightDecayRegularizer):
             main_prog = fluid.Program()
             startup_prog = fluid.Program()
             with fluid.program_guard(main_prog, startup_prog):
-                data = fluid.layers.data(name='image', shape=[3, 28, 28], dtype='float32')
-                label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+                data = paddle.static.data(name='image', shape=[-1, 3, 28, 28], dtype='float32')
+                label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
                 hidden = paddle.static.nn.fc(x=data, size=128, activation='relu')
                 prediction = paddle.static.nn.fc(x=hidden, size=10, activation='softmax')
                 loss = paddle.nn.functional.cross_entropy(
@@ -193,8 +193,8 @@ class L1DecayRegularizer(WeightDecayRegularizer):
             main_prog = fluid.Program()
             startup_prog = fluid.Program()
             with fluid.program_guard(main_prog, startup_prog):
-                data = fluid.layers.data(name='image', shape=[3, 28, 28], dtype='float32')
-                label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+                data = paddle.static.data(name='image', shape=[-1, 3, 28, 28], dtype='float32')
+                label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
                 hidden = paddle.static.nn.fc(x=data, size=128, activation='relu')
                 prediction = paddle.static.nn.fc(x=hidden, size=10, activation='softmax')
                 loss = paddle.nn.functional.cross_entropy(
diff --git a/python/paddle/fluid/tests/book/notest_understand_sentiment.py b/python/paddle/fluid/tests/book/notest_understand_sentiment.py
index c406fae5f811c8edf9dc45209f2e576663e35832..0eb09b210bd02962f75b48a891ac28a0c6610d72 100644
--- a/python/paddle/fluid/tests/book/notest_understand_sentiment.py
+++ b/python/paddle/fluid/tests/book/notest_understand_sentiment.py
@@ -68,10 +68,10 @@ def train(
     dict_dim = len(word_dict)
     class_dim = 2
 
-    data = fluid.layers.data(
-        name="words", shape=[1], dtype="int64", lod_level=1
+    data = paddle.static.data(
+        name="words", shape=[-1, 1], dtype="int64", lod_level=1
     )
-    label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+    label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
 
     if not parallel:
         cost, acc_out, prediction = net_method(
diff --git a/python/paddle/fluid/tests/book/test_fit_a_line.py b/python/paddle/fluid/tests/book/test_fit_a_line.py
index b6efc9775efa4572afd1706441ee085345b74aef..c7c930fd53b5880eb5c83fff5777ac9d1e7db5d2 100644
--- a/python/paddle/fluid/tests/book/test_fit_a_line.py
+++ b/python/paddle/fluid/tests/book/test_fit_a_line.py
@@ -49,8 +49,10 @@ def convert_float_to_uint16(in_list):
 
 
 def train(use_cuda, save_dirname, is_local, use_bf16, pure_bf16):
-    x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-    y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+    x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+    x.desc.set_need_check_feed(False)
+    y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
+    y.desc.set_need_check_feed(False)
 
     if use_bf16:
         if not pure_bf16:
diff --git a/python/paddle/fluid/tests/book/test_image_classification.py b/python/paddle/fluid/tests/book/test_image_classification.py
index f1fa47afb809ab3e247cf15d7d2d9379b2185340..6ee33951583a1cf2be7acaac7fa49d71c71a5c49 100644
--- a/python/paddle/fluid/tests/book/test_image_classification.py
+++ b/python/paddle/fluid/tests/book/test_image_classification.py
@@ -104,8 +104,10 @@ def train(net_type, use_cuda, save_dirname, is_local):
     classdim = 10
     data_shape = [3, 32, 32]
 
-    images = fluid.layers.data(name='pixel', shape=data_shape, dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    images = paddle.static.data(
+        name='pixel', shape=[-1] + data_shape, dtype='float32'
+    )
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     if net_type == "vgg":
         print("train vgg net")
diff --git a/python/paddle/fluid/tests/book/test_recognize_digits.py b/python/paddle/fluid/tests/book/test_recognize_digits.py
index 85d946e2185846b44fcd852842343eab04020680..5d6939af8902fdca51b6b4486788269194524b6b 100644
--- a/python/paddle/fluid/tests/book/test_recognize_digits.py
+++ b/python/paddle/fluid/tests/book/test_recognize_digits.py
@@ -77,8 +77,8 @@ def train(
 ):
     if use_cuda and not fluid.core.is_compiled_with_cuda():
         return
-    img = fluid.layers.data(name='img', shape=[1, 28, 28], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='img', shape=[-1, 1, 28, 28], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     if nn_type == 'mlp':
         net_conf = mlp
diff --git a/python/paddle/fluid/tests/book/test_recommender_system.py b/python/paddle/fluid/tests/book/test_recommender_system.py
index aad9e7ce01c2c0dc654fcd7cfc09d3521c4aa1bb..1564b0594f2a12b25378ccab748bd6230ca805e6 100644
--- a/python/paddle/fluid/tests/book/test_recommender_system.py
+++ b/python/paddle/fluid/tests/book/test_recommender_system.py
@@ -40,7 +40,7 @@ def get_usr_combined_features():
 
     USR_DICT_SIZE = paddle.dataset.movielens.max_user_id() + 1
 
-    uid = layers.data(name='user_id', shape=[1], dtype='int64')
+    uid = paddle.static.data(name='user_id', shape=[-1, 1], dtype='int64')
 
     usr_emb = layers.embedding(
         input=uid,
@@ -54,7 +54,9 @@ def get_usr_combined_features():
 
     USR_GENDER_DICT_SIZE = 2
 
-    usr_gender_id = layers.data(name='gender_id', shape=[1], dtype='int64')
+    usr_gender_id = paddle.static.data(
+        name='gender_id', shape=[-1, 1], dtype='int64'
+    )
 
     usr_gender_emb = layers.embedding(
         input=usr_gender_id,
@@ -66,7 +68,7 @@ def get_usr_combined_features():
     usr_gender_fc = paddle.static.nn.fc(x=usr_gender_emb, size=16)
 
     USR_AGE_DICT_SIZE = len(paddle.dataset.movielens.age_table)
-    usr_age_id = layers.data(name='age_id', shape=[1], dtype="int64")
+    usr_age_id = paddle.static.data(name='age_id', shape=[-1, 1], dtype="int64")
 
     usr_age_emb = layers.embedding(
         input=usr_age_id,
@@ -78,7 +80,7 @@ def get_usr_combined_features():
     usr_age_fc = paddle.static.nn.fc(x=usr_age_emb, size=16)
 
     USR_JOB_DICT_SIZE = paddle.dataset.movielens.max_job_id() + 1
-    usr_job_id = layers.data(name='job_id', shape=[1], dtype="int64")
+    usr_job_id = paddle.static.data(name='job_id', shape=[-1, 1], dtype="int64")
 
     usr_job_emb = layers.embedding(
         input=usr_job_id,
@@ -104,7 +106,7 @@ def get_mov_combined_features():
 
     MOV_DICT_SIZE = paddle.dataset.movielens.max_movie_id() + 1
 
-    mov_id = layers.data(name='movie_id', shape=[1], dtype='int64')
+    mov_id = paddle.static.data(name='movie_id', shape=[-1, 1], dtype='int64')
 
     mov_emb = layers.embedding(
         input=mov_id,
@@ -118,8 +120,8 @@ def get_mov_combined_features():
 
     CATEGORY_DICT_SIZE = len(paddle.dataset.movielens.movie_categories())
 
-    category_id = layers.data(
-        name='category_id', shape=[1], dtype='int64', lod_level=1
+    category_id = paddle.static.data(
+        name='category_id', shape=[-1, 1], dtype='int64', lod_level=1
     )
 
     mov_categories_emb = layers.embedding(
@@ -132,8 +134,8 @@ def get_mov_combined_features():
 
     MOV_TITLE_DICT_SIZE = len(paddle.dataset.movielens.get_movie_title_dict())
 
-    mov_title_id = layers.data(
-        name='movie_title', shape=[1], dtype='int64', lod_level=1
+    mov_title_id = paddle.static.data(
+        name='movie_title', shape=[-1, 1], dtype='int64', lod_level=1
     )
 
     mov_title_emb = layers.embedding(
@@ -170,7 +172,7 @@ def model():
     )
     scale_infer = paddle.scale(x=inference, scale=5.0)
 
-    label = layers.data(name='score', shape=[1], dtype='float32')
+    label = paddle.static.data(name='score', shape=[-1, 1], dtype='float32')
     square_cost = paddle.nn.functional.square_error_cost(
         input=scale_infer, label=label
     )
diff --git a/python/paddle/fluid/tests/book/test_word2vec_book.py b/python/paddle/fluid/tests/book/test_word2vec_book.py
index e932394e8cc0952f417d2d155c170c9d0d3a01d2..f117c523a74444570630b6ed06763c906ea87b2a 100644
--- a/python/paddle/fluid/tests/book/test_word2vec_book.py
+++ b/python/paddle/fluid/tests/book/test_word2vec_book.py
@@ -108,11 +108,13 @@ def train(
     word_dict = paddle.dataset.imikolov.build_dict()
     dict_size = len(word_dict)
 
-    first_word = fluid.layers.data(name='firstw', shape=[1], dtype='int64')
-    second_word = fluid.layers.data(name='secondw', shape=[1], dtype='int64')
-    third_word = fluid.layers.data(name='thirdw', shape=[1], dtype='int64')
-    forth_word = fluid.layers.data(name='forthw', shape=[1], dtype='int64')
-    next_word = fluid.layers.data(name='nextw', shape=[1], dtype='int64')
+    first_word = paddle.static.data(name='firstw', shape=[-1, 1], dtype='int64')
+    second_word = paddle.static.data(
+        name='secondw', shape=[-1, 1], dtype='int64'
+    )
+    third_word = paddle.static.data(name='thirdw', shape=[-1, 1], dtype='int64')
+    forth_word = paddle.static.data(name='forthw', shape=[-1, 1], dtype='int64')
+    next_word = paddle.static.data(name='nextw', shape=[-1, 1], dtype='int64')
 
     if not is_parallel:
         avg_cost, predict_word = __network__(
diff --git a/python/paddle/fluid/tests/test_data_feeder.py b/python/paddle/fluid/tests/test_data_feeder.py
index 4cbdba6419501cdba3ec671eb055bd04f4e1537c..517d874f4544efdb6023d9d94715b6b8ce52c64d 100644
--- a/python/paddle/fluid/tests/test_data_feeder.py
+++ b/python/paddle/fluid/tests/test_data_feeder.py
@@ -22,8 +22,8 @@ paddle.enable_static()
 
 class TestDataFeeder(unittest.TestCase):
     def test_lod_level_0_converter(self):
-        img = fluid.layers.data(name='image', shape=[1, 28, 28])
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        img = paddle.static.data(name='image', shape=[-1, 1, 28, 28])
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         feeder = fluid.DataFeeder([img, label], fluid.CPUPlace())
         result = feeder.feed([([0] * 784, [9]), ([1] * 784, [1])])
 
@@ -41,10 +41,10 @@ class TestDataFeeder(unittest.TestCase):
     def test_lod_level_1_converter(self):
         # lod_level = 1
         # each sentence has a different number of words
-        sentences = fluid.layers.data(
-            name='sentences', shape=[1], dtype='int64', lod_level=1
+        sentences = paddle.static.data(
+            name='sentences', shape=[-1, 1], dtype='int64', lod_level=1
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         feeder = fluid.DataFeeder([sentences, label], fluid.CPUPlace())
 
         # lod = [[0, 3, 5, 9]]
@@ -64,10 +64,10 @@ class TestDataFeeder(unittest.TestCase):
     def test_lod_level_2_converter(self):
         # lod_level = 2
         # paragraphs -> sentences -> words
-        paragraphs = fluid.layers.data(
-            name='paragraphs', shape=[1], dtype='int64', lod_level=2
+        paragraphs = paddle.static.data(
+            name='paragraphs', shape=[-1, 1], dtype='int64', lod_level=2
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         feeder = fluid.DataFeeder([paragraphs, label], fluid.CPUPlace())
 
         # lod = [[0, 2, 3], [0, 3, 5, 9]]
diff --git a/python/paddle/fluid/tests/test_detection.py b/python/paddle/fluid/tests/test_detection.py
index aa97a320549290b21a7c54c506d0ee966793b307..39b17cd0634be23bd81342e178087bb1815dbfe6 100644
--- a/python/paddle/fluid/tests/test_detection.py
+++ b/python/paddle/fluid/tests/test_detection.py
@@ -20,7 +20,6 @@ from unittests.test_imperative_base import new_program_scope
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 from paddle.fluid.dygraph import base
 from paddle.fluid.framework import Program, program_guard
@@ -154,10 +153,12 @@ class TestMulticlassNMS2(unittest.TestCase):
     def test_multiclass_nms2(self):
         program = Program()
         with program_guard(program):
-            bboxes = layers.data(
+            bboxes = paddle.static.data(
                 name='bboxes', shape=[-1, 10, 4], dtype='float32'
             )
-            scores = layers.data(name='scores', shape=[-1, 10], dtype='float32')
+            scores = paddle.static.data(
+                name='scores', shape=[-1, 10], dtype='float32'
+            )
             output = fluid.contrib.multiclass_nms2(
                 bboxes, scores, 0.3, 400, 200, 0.7
             )
diff --git a/python/paddle/fluid/tests/test_error_clip.py b/python/paddle/fluid/tests/test_error_clip.py
index 9dc0771a39fb868792f17dd5497ee7031804147e..0c1d8614938cd4f5e97ae132966f7f33e13009df 100644
--- a/python/paddle/fluid/tests/test_error_clip.py
+++ b/python/paddle/fluid/tests/test_error_clip.py
@@ -23,13 +23,13 @@ paddle.enable_static()
 prog = fluid.framework.Program()
 
 with fluid.program_guard(main_program=prog):
-    image = fluid.layers.data(name='x', shape=[784], dtype='float32')
+    image = paddle.static.data(name='x', shape=[-1, 784], dtype='float32')
 
     hidden1 = paddle.static.nn.fc(x=image, size=128, activation='relu')
     hidden2 = paddle.static.nn.fc(x=hidden1, size=64, activation='relu')
     predict = paddle.static.nn.fc(x=hidden2, size=10, activation='softmax')
 
-    label = fluid.layers.data(name='y', shape=[1], dtype='int64')
+    label = paddle.static.data(name='y', shape=[-1, 1], dtype='int64')
 
     cost = paddle.nn.functional.cross_entropy(
         input=predict, label=label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/check_nan_inf_base.py b/python/paddle/fluid/tests/unittests/check_nan_inf_base.py
index 551a9eed8baa8f4795801a01c70e8363c2e9830e..954e06e2a81f05d654fc19b41cb1ec8342760c22 100644
--- a/python/paddle/fluid/tests/unittests/check_nan_inf_base.py
+++ b/python/paddle/fluid/tests/unittests/check_nan_inf_base.py
@@ -46,8 +46,8 @@ def generator():
 
 
 def net():
-    x = fluid.layers.data(name="x", shape=[3], dtype='float32')
-    y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+    x = paddle.static.data(name="x", shape=[-1, 3], dtype='float32')
+    y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
     # test int64 value
     zero = fluid.layers.fill_constant(shape=[1], dtype='int64', value=0)
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_allgather_api.py b/python/paddle/fluid/tests/unittests/collective/collective_allgather_api.py
index 72e28c21f250c11e74522b55a92d89db5d3891cd..6eb1f562860baee24948ad259ebd3eec15525c20 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_allgather_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_allgather_api.py
@@ -20,7 +20,6 @@ import test_collective_api_base as test_base
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -33,7 +32,9 @@ class TestCollectiveAllgatherAPI(test_base.TestCollectiveAPIRunnerBase):
         dtype = "float32" if dtype is None else dtype
         with fluid.program_guard(main_prog, startup_program):
             tensor_list = []
-            tindata = layers.data(name="tindata", shape=[10, 1000], dtype=dtype)
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype=dtype
+            )
             paddle.distributed.all_gather(tensor_list, tindata)
             return tensor_list
 
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_api.py b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_api.py
index e995cd8d37b6623cbb6f7ef2739d21b545e8a46d..291ad384f3e66e357661eed02a0c1bd406b4d365 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_api.py
@@ -16,7 +16,6 @@ from test_collective_api_base import TestCollectiveAPIRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -27,8 +26,8 @@ class TestCollectiveAllreduceAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             paddle.distributed.all_reduce(tindata)
             return [tindata]
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_new_group_api.py b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_new_group_api.py
index ba5580eb28645c78a5ffe3a5ccc189d1aea1cd88..b66fdbedd2e0469a5d6fb8b1452d7bc1c1c8ecd5 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_new_group_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_new_group_api.py
@@ -16,7 +16,6 @@ from test_collective_api_base import TestCollectiveAPIRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -27,8 +26,8 @@ class TestCollectiveAllreduceNewGroupAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[1, 10, 1000], dtype='float32'
             )
             gp = paddle.distributed.new_group([0, 1])
             paddle.distributed.all_reduce(tindata, group=gp, sync_op=True)
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op.py b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op.py
index 3f3d0e8c0a639767e5972b489ef150b784e6acb4..250156905e024188a871b6a54633e44c19d6238b 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -29,9 +28,10 @@ class TestCollectiveAllreduce(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program):
         ring_id = 0
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofallreduce",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op_wait.py b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op_wait.py
index 0fce303f784c6d0496e42c902cbbde4dec2085e8..b0d4eeec950993a7b193549eb76fab6b23f40f62 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op_wait.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_allreduce_op_wait.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -29,9 +28,10 @@ class TestCollectiveAllreduce(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program):
         ring_id = 0
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofallreduce",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_alltoall_api.py b/python/paddle/fluid/tests/unittests/collective/collective_alltoall_api.py
index 0bfc3951f9146ede7364b613916481a3d076ffcc..1b5d0082ead7cfeb44c565007f07522e810eef5b 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_alltoall_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_alltoall_api.py
@@ -16,7 +16,6 @@ from test_collective_api_base import TestCollectiveAPIRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -27,9 +26,10 @@ class TestCollectiveAllToAllAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             tindata = paddle.split(tindata, 2, axis=0)
             tout_data = []
             paddle.distributed.alltoall(tindata, tout_data)
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_broadcast_api.py b/python/paddle/fluid/tests/unittests/collective/collective_broadcast_api.py
index 046501ca205fdc6fa1c2b99a324e89f8c310db4e..5011a5b79eba692eeb6b44880b50667d04111e96 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_broadcast_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_broadcast_api.py
@@ -16,7 +16,6 @@ from test_collective_api_base import TestCollectiveAPIRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -27,9 +26,10 @@ class TestCollectiveBroadcastAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             paddle.distributed.broadcast(tindata, src=1)
             return [tindata]
 
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_broadcast_op.py b/python/paddle/fluid/tests/unittests/collective/collective_broadcast_op.py
index 8d02dd771bfdbdcb6bc121a7f62411cce6d12ae6..7c49f1f55ebe6975a10ee03809ae98f51337fc1b 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_broadcast_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_broadcast_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveBroadcast(TestCollectiveRunnerBase):
         ring_id = 0
         rootid = 1
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofbroadcast",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_concat_op.py b/python/paddle/fluid/tests/unittests/collective/collective_concat_op.py
index dd85905caf288c0f2d315163355713e62527df45..df784d2815eeeae338bb255784f465b21061c105 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_concat_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_concat_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveConcat(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofconcat",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_identity_op.py b/python/paddle/fluid/tests/unittests/collective/collective_identity_op.py
index c4b993d8a311fb6d3e76769f0fd5a996cfdde0fc..4926076b6b0b3db77cbf527fdb9ee1beff80cca0 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_identity_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_identity_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveIdentity(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofgather",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_reduce_api.py b/python/paddle/fluid/tests/unittests/collective/collective_reduce_api.py
index fb16088b4b87fafe797038266f59e76804aa3eea..6f033c7d1fdeb24f9b46ee6470d451cf0adfd2fa 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_reduce_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_reduce_api.py
@@ -16,7 +16,6 @@ from test_collective_api_base import TestCollectiveAPIRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -27,9 +26,10 @@ class TestCollectiveReduceAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             paddle.distributed.reduce(tindata, dst=0)
             return [tindata]
 
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_reduce_op.py b/python/paddle/fluid/tests/unittests/collective/collective_reduce_op.py
index 8e75d5ee913020db9c824542db0852a2b0987392..9c340821130910d29bd8aa69376c2305e9b0c50f 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_reduce_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_reduce_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveReduce(TestCollectiveRunnerBase):
         ring_id = 0
         rootid = 1
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofreduce",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_reduce_op_calc_stream.py b/python/paddle/fluid/tests/unittests/collective/collective_reduce_op_calc_stream.py
index 7d122764f5d4da6b9a094faa4bbc9760e36c3774..a4ae2a3623f60bdfcac94f9470418fdccb8b8ed2 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_reduce_op_calc_stream.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_reduce_op_calc_stream.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,11 @@ class TestCollectiveReduce(TestCollectiveRunnerBase):
         ring_id = 0
         rootid = 1
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
+
             toutdata = main_prog.current_block().create_var(
                 name="outofreduce",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_scatter_api.py b/python/paddle/fluid/tests/unittests/collective/collective_scatter_api.py
index f4671af0dfc7f09ba2fa973c61637860b56ccec6..de7ec2cfcd377d7495a63e41e3a899f763100c57 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_scatter_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_scatter_api.py
@@ -27,11 +27,10 @@ class TestCollectiveScatterAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
+            tindata = paddle.static.data(
                 name="tindata",
                 shape=[10, 1000],
                 dtype='float32',
-                append_batch_size=False,
             )
             toutdata = layers.fill_constant(
                 shape=[5, 1000], dtype='float32', value=1.0
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_scatter_op.py b/python/paddle/fluid/tests/unittests/collective/collective_scatter_op.py
index 798485b638d40359a2937f37ffe5a9aec121ac82..56aa7210d27606b3bd23933c47ba63e7dd280e7b 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_scatter_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_scatter_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveScatter(TestCollectiveRunnerBase):
         ring_id = 0
         rootid = 1
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofreduce",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_api.py b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_api.py
index bcdb8342cadd28e80709c623c65b79564dd18d49..e0a2487743d4b51ef0e016b1bd7bf5d1c69f5b7e 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_api.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_api.py
@@ -16,7 +16,6 @@ from test_collective_api_base import TestCollectiveAPIRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -27,11 +26,10 @@ class TestCollectiveSendRecvAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
+            tindata = paddle.static.data(
                 name="tindata",
                 shape=[10, 1000],
                 dtype='float32',
-                append_batch_size=False,
             )
             if rank == 0:
                 paddle.distributed.send(tindata, dst=1)
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op.py b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op.py
index eab498073323e49cbe84871cbf78dea8122c7535..8b2abc74f0aaf064f06ee58fc66df6b4212eb3cd 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -28,12 +27,12 @@ class TestCollectiveSendRecv(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program):
         ring_id = self.global_ring_id
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
+            tindata = paddle.static.data(
                 name="tindata",
                 shape=[10, 1000],
                 dtype='float64',
-                append_batch_size=False,
             )
+            tindata.desc.set_need_check_feed(False)
             if self.rank == 0:
                 main_prog.global_block().append_op(
                     type="send_v2",
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_array.py b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_array.py
index 8c4ebeaffda80b0d267379751dbf0125decafc83..bea4a71089af42c1412a9e1eacd9a59a067d7db0 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_array.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_array.py
@@ -17,7 +17,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -29,12 +28,12 @@ class TestCollectiveSendRecv(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program):
         ring_id = self.global_ring_id
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
+            tindata = paddle.static.data(
                 name="tindata",
                 shape=[10, 1000],
                 dtype='float64',
-                append_batch_size=False,
             )
+            tindata.desc.set_need_check_feed(False)
             if self.rank == 0:
                 data1 = fluid.layers.assign(
                     np.array([[0, 1, 2]], dtype='float32')
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_dynamic_shape.py b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_dynamic_shape.py
index 80a0176782c4ec16246baa749b3cdc6b0190aaf7..2f1aaac0b19a1b9f26ea920c3c9e8bae5b336d48 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_dynamic_shape.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_sendrecv_op_dynamic_shape.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -28,12 +27,12 @@ class TestCollectiveSendRecvDynamicShape(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program):
         ring_id = self.global_ring_id
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
+            tindata = paddle.static.data(
                 name="tindata",
-                shape=[10, 1000],
+                shape=[-1, 10, 1000],
                 dtype='float64',
-                append_batch_size=False,
             )
+            tindata.desc.set_need_check_feed(False)
             if self.rank == 0:
                 main_prog.global_block().append_op(
                     type="send_v2",
diff --git a/python/paddle/fluid/tests/unittests/collective/collective_split_op.py b/python/paddle/fluid/tests/unittests/collective/collective_split_op.py
index 127ab924da4f5bbb42fe15d6cf3e1bff08e47998..3348d4c9c9f29d02f538484021209cd7187c0533 100644
--- a/python/paddle/fluid/tests/unittests/collective/collective_split_op.py
+++ b/python/paddle/fluid/tests/unittests/collective/collective_split_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveAllGather(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofsplit",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/dist_mnist_gradient_merge.py b/python/paddle/fluid/tests/unittests/collective/fleet/dist_mnist_gradient_merge.py
index 01d6d970c6cd2c8f4f54ce8ede7576fb514f1bc4..35525d03ca7a6ba099a149acff36cce467ff42f1 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/dist_mnist_gradient_merge.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/dist_mnist_gradient_merge.py
@@ -29,8 +29,10 @@ fluid.default_main_program().random_seed = 1
 class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist.py b/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist.py
index c33e5bc27a0cae34f75c223aa305c8a612041a66..2eb0951756a59eaa601549e3974872110e1ad4a4 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist.py
@@ -85,10 +85,12 @@ class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2, use_dgc=False, dist_strategy=None):
         # Input data
         with fluid.device_guard("gpu:0"):
-            images = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype=DTYPE
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
 
             if dist_strategy:
                 data_loader = fluid.io.DataLoader.from_generator(
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_multi_device.py b/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_multi_device.py
index 905df8cd6b71f82337c8a50220ab20d66bc8b74a..e094d932d33e48c2a7d51a6c30859cebad6aadf6 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_multi_device.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_multi_device.py
@@ -85,10 +85,12 @@ class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2, use_dgc=False, dist_strategy=None):
         # Input data
         with fluid.device_guard("gpu:0"):
-            images = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype=DTYPE
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
 
             if dist_strategy:
                 data_loader = fluid.io.DataLoader.from_generator(
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_one_device.py b/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_one_device.py
index dbe50789d687c1c92003ee7ac750bbd2cb455a78..7e442f1914b2cf05d6f4f8f71856c02add563991 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_one_device.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/pipeline_mnist_one_device.py
@@ -77,10 +77,12 @@ class TestDistMnist2x2(TestDistRunnerBase):
         if dist_strategy:
             fleet.init(is_collective=True)
         with fluid.device_guard("gpu:0"):
-            images = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype=DTYPE
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
 
             if dist_strategy:
                 data_loader = fluid.io.DataLoader.from_generator(
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_half_async.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_half_async.py
index 21f7b624f5a4dfb0bd24f6cb069f6b723f335c55..6d8decbc3e998647940ba12e26de98f472216fc4 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_half_async.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_half_async.py
@@ -29,9 +29,9 @@ paddle.enable_static()
 
 class TestCommunicatorHalfAsyncEnd2End(unittest.TestCase):
     def net(self):
-        x = fluid.layers.data(name='x', shape=[13], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
         y_predict = paddle.static.nn.fc(x, size=1, activation=None)
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_sync.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_sync.py
index 550fc5db902416281c30095f089cdf19e9fa0b81..cf0babc5877333bf10ca92fd86278a0719b6412a 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_sync.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_communicator_sync.py
@@ -27,8 +27,8 @@ import paddle.fluid as fluid
 
 class TestCommunicator(unittest.TestCase):
     def net(self):
-        x = fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
         return avg_cost
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_distributed_strategy.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_distributed_strategy.py
index 3e58391ec976882a05f009c3b93bb031e2cbe30b..085cb293c0e46418fea8cf73850dc257d8e2aca0 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_distributed_strategy.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_distributed_strategy.py
@@ -270,8 +270,8 @@ class TestHalfAsyncStrategy(unittest.TestCase):
 
 class TestDebugInfo(unittest.TestCase):
     def test_debug_info(self):
-        x = fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         y_predict = paddle.static.nn.fc(x, size=1, activation=None)
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_fp16_allreduce_meta_optimizer.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_fp16_allreduce_meta_optimizer.py
index 0e5ae267f32af44119e0de912b50e22f1e8f3bf7..84dd2e0ee964b5f4dd1c8ca903e12097f583e51b 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_fp16_allreduce_meta_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_fp16_allreduce_meta_optimizer.py
@@ -30,12 +30,8 @@ class TestFleetFP16CompressOptimizer(unittest.TestCase):
 
     def net(self, main_prog, startup_prog, dtype='float32'):
         with fluid.program_guard(main_prog, startup_prog):
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype=dtype
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
-            )
+            input_x = paddle.static.data(name="x", shape=[-1, 32], dtype=dtype)
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_execution_meta_optimizer.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_execution_meta_optimizer.py
index aa8bc8bd2d73989600def09d8bc830924fa375bd..a36c5a1d74cc08e3e1d290cd38068f2e4fdd0273 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_execution_meta_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_execution_meta_optimizer.py
@@ -60,12 +60,10 @@ class TestFleetGraphExecutionMetaOptimizer(unittest.TestCase):
             import paddle.distributed.fleet as fleet
 
             fleet.init(is_collective=True)
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
@@ -126,12 +124,10 @@ class TestFleetGraphExecutionMetaOptimizer(unittest.TestCase):
             import paddle.distributed.fleet as fleet
 
             fleet.init(is_collective=True)
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
@@ -204,12 +200,10 @@ class TestFleetGraphExecutionMetaOptimizer(unittest.TestCase):
             import paddle.distributed.fleet as fleet
 
             fleet.init(is_collective=True)
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
@@ -269,12 +263,10 @@ class TestFleetGraphExecutionMetaOptimizer(unittest.TestCase):
             import paddle.distributed.fleet as fleet
 
             fleet.init(is_collective=True)
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_executor.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_executor.py
index 58091bd847f2c960a58b32da9ffb40b7755d1608..aab4032afbce7efeab1c63b4b6bdc3a885453d5d 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_executor.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_graph_executor.py
@@ -45,12 +45,10 @@ class TestFleetGraphExecutionMetaOptimizer(unittest.TestCase):
         def node_func():
             role = role_maker.PaddleCloudRoleMaker(is_collective=True)
             fleet.init(role)
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lamb_meta_optimizer.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lamb_meta_optimizer.py
index 2623a222d50b42d2c8c6bd68f8a8e8d3b54faca7..96cf0fffe87be160e51dc4570c94b0cb484d50d5 100755
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lamb_meta_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lamb_meta_optimizer.py
@@ -33,11 +33,11 @@ class TestFleetLambMetaOptimizer(unittest.TestCase):
     def net(self, main_prog, startup_prog):
         with fluid.program_guard(main_prog, startup_prog):
             with fluid.unique_name.guard():
-                input_x = paddle.fluid.layers.data(
-                    name="x", shape=[32], dtype='float32'
+                input_x = paddle.static.data(
+                    name="x", shape=[-1, 32], dtype='float32'
                 )
-                input_y = paddle.fluid.layers.data(
-                    name="y", shape=[1], dtype='int64'
+                input_y = paddle.static.data(
+                    name="y", shape=[-1, 1], dtype='int64'
                 )
 
                 fc_1 = paddle.static.nn.fc(
@@ -117,10 +117,8 @@ class TestFleetLambMetaOptimizer(unittest.TestCase):
     def test_lamb_apply_with_amp(self):
         role = role_maker.PaddleCloudRoleMaker(is_collective=True)
         fleet.init(role)
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lars_meta_optimizer.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lars_meta_optimizer.py
index 509606290144606691a74a93a0877e64c8d374fb..b3094b3b6b0311506328f4d165e6db651bfc5ff1 100755
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lars_meta_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_lars_meta_optimizer.py
@@ -33,11 +33,11 @@ class TestFleetLarsMetaOptimizer(unittest.TestCase):
     def net(self, main_prog, startup_prog):
         with fluid.program_guard(main_prog, startup_prog):
             with fluid.unique_name.guard():
-                input_x = paddle.fluid.layers.data(
-                    name="x", shape=[32], dtype='float32'
+                input_x = paddle.static.data(
+                    name="x", shape=[-1, 32], dtype='float32'
                 )
-                input_y = paddle.fluid.layers.data(
-                    name="y", shape=[1], dtype='int64'
+                input_y = paddle.static.data(
+                    name="y", shape=[-1, 1], dtype='int64'
                 )
 
                 fc_1 = paddle.static.nn.fc(
@@ -122,10 +122,8 @@ class TestFleetLarsMetaOptimizer(unittest.TestCase):
     def test_lars_apply_with_amp(self):
         role = role_maker.PaddleCloudRoleMaker(is_collective=True)
         fleet.init(role)
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_meta_optimizer_base.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_meta_optimizer_base.py
index 76fb129a0494617e3e7c04018756aa88891088c6..3551c89a2601487b92095ae16d1a3d16b063801a 100755
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_meta_optimizer_base.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_meta_optimizer_base.py
@@ -29,11 +29,11 @@ class TestFleetMetaOptimizerBase(unittest.TestCase):
             with fluid.unique_name.guard():
                 role = role_maker.PaddleCloudRoleMaker(is_collective=True)
                 fleet.init(role)
-                input_x = paddle.fluid.layers.data(
-                    name="x", shape=[32], dtype='float32'
+                input_x = paddle.static.data(
+                    name="x", shape=[-1, 32], dtype='float32'
                 )
-                input_y = paddle.fluid.layers.data(
-                    name="y", shape=[1], dtype='int64'
+                input_y = paddle.static.data(
+                    name="y", shape=[-1, 1], dtype='int64'
                 )
 
                 fc_1 = paddle.static.nn.fc(
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer.py
index 8fac45b9d249a951a68050759d4821868d7ae5da..ddba6331998264964ad94024d73e6b6b2a14296f 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer.py
@@ -33,14 +33,12 @@ class TestFleetMetaOptimizer(unittest.TestCase):
 
     def net(self):
         with static.device_guard("gpu:0"):
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
-            )
-            input_z = paddle.fluid.layers.data(
-                name="z", shape=[1], dtype="float32"
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
+            input_z = paddle.static.data(
+                name="z", shape=[-1, 1], dtype="float32"
             )
             with static.device_guard("gpu:all"):
                 input_z = input_z * 1.0
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer_with_recompute.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer_with_recompute.py
index 3f22238a3637c474dc224f01fd77c0e262d3f76b..7c80ad98597aa81b101075cbc3f7eefc1ad5a3d9 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer_with_recompute.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_pipeline_meta_optimizer_with_recompute.py
@@ -34,12 +34,10 @@ class TestFleetMetaOptimizer(unittest.TestCase):
         role = role_maker.PaddleCloudRoleMaker(is_collective=True)
         fleet.init(role)
         with paddle.fluid.device_guard("gpu:0"):
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
             fc_3 = paddle.static.nn.fc(x=fc_2, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_raw_program_meta_optimizer.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_raw_program_meta_optimizer.py
index 29ef57518d0c5308071f95b6b2135f4c4269796d..62b3ba99eaf8d5c2e9517b2651b06604b8078e06 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_raw_program_meta_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_raw_program_meta_optimizer.py
@@ -33,10 +33,8 @@ class TestFleetMetaOptimizer(unittest.TestCase):
 
         role = role_maker.PaddleCloudRoleMaker(is_collective=True)
         fleet.init(role)
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
 
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_rolemaker_new.py b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_rolemaker_new.py
index 196e3425ec65a8c7ce5f0e623bd6042ea8f79926..7e687845944647110049b60937d96ae7f3edbb9c 100644
--- a/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_rolemaker_new.py
+++ b/python/paddle/fluid/tests/unittests/collective/fleet/test_fleet_rolemaker_new.py
@@ -443,9 +443,9 @@ class TestGlooWithCloudRoleMaker(unittest.TestCase):
         os.environ["PADDLE_GLOO_FS_PATH"] = tmp
 
         def net():
-            x = paddle.fluid.layers.data(name='x', shape=[13], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
-            y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
             )
diff --git a/python/paddle/fluid/tests/unittests/collective_allgather_op.py b/python/paddle/fluid/tests/unittests/collective_allgather_op.py
index e877aaae9bd8f143041e0938dadc1a2f357842ff..5fd37ebe93878c24e7ebecf713e2adc2a8d51bad 100644
--- a/python/paddle/fluid/tests/unittests/collective_allgather_op.py
+++ b/python/paddle/fluid/tests/unittests/collective_allgather_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveAllGather(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofgather",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/collective_reducescatter.py b/python/paddle/fluid/tests/unittests/collective_reducescatter.py
index 0852f7b2f36628d00bc390f05eb062d7ef7a0abb..9813553295fbcdee182d5948c2647f2ebce91451 100644
--- a/python/paddle/fluid/tests/unittests/collective_reducescatter.py
+++ b/python/paddle/fluid/tests/unittests/collective_reducescatter.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -29,9 +28,10 @@ class TestCollectiveReduceScatter(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = fluid.layers.collective._c_reducescatter(tindata, nranks)
             toutdata = fluid.layers.collective._c_sync_comm_stream(toutdata, 0)
             return toutdata
diff --git a/python/paddle/fluid/tests/unittests/collective_reducescatter_op.py b/python/paddle/fluid/tests/unittests/collective_reducescatter_op.py
index d89701e6a4a802e3b57841c1f5ce85dc6b4016b5..92c9ec3b6ed6ad5db5b1f3e03929e368755b6f70 100644
--- a/python/paddle/fluid/tests/unittests/collective_reducescatter_op.py
+++ b/python/paddle/fluid/tests/unittests/collective_reducescatter_op.py
@@ -16,7 +16,6 @@ from test_collective_base import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,9 +29,10 @@ class TestCollectiveReduceScatter(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofrs",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/dist_allreduce_op.py b/python/paddle/fluid/tests/unittests/dist_allreduce_op.py
index 70d06a95b64aaeee9b95a0a6b06966fc45d47328..30bcea4cb5cb2344f3d50e3d10d1951dbedabc71 100644
--- a/python/paddle/fluid/tests/unittests/dist_allreduce_op.py
+++ b/python/paddle/fluid/tests/unittests/dist_allreduce_op.py
@@ -72,8 +72,10 @@ def cnn_model(data):
 class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2, single_device=False):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_ctr.py b/python/paddle/fluid/tests/unittests/dist_ctr.py
index 7fd86cadb99eb733c5f37d33725b42e4568282e0..deb4cb921c1f3a4809e3d97dd1d7a0b0a78d5251 100644
--- a/python/paddle/fluid/tests/unittests/dist_ctr.py
+++ b/python/paddle/fluid/tests/unittests/dist_ctr.py
@@ -33,26 +33,23 @@ class TestDistCTR2x2(TestDistRunnerBase):
 
         dnn_input_dim, lr_input_dim = dist_ctr_reader.load_data_meta()
         """ network definition """
-        dnn_data = fluid.layers.data(
+        dnn_data = paddle.static.data(
             name="dnn_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        lr_data = fluid.layers.data(
+        lr_data = paddle.static.data(
             name="lr_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        label = fluid.layers.data(
+        label = paddle.static.data(
             name="click",
             shape=[-1, 1],
             dtype="int64",
             lod_level=0,
-            append_batch_size=False,
         )
 
         # build dnn model
diff --git a/python/paddle/fluid/tests/unittests/dist_fleet_ctr.py b/python/paddle/fluid/tests/unittests/dist_fleet_ctr.py
index 360cad434096be3e4705dc9b8367f3c9f6089bf6..8e9341f9c5b1a6b27d9e7c447e35cddcca71e13a 100644
--- a/python/paddle/fluid/tests/unittests/dist_fleet_ctr.py
+++ b/python/paddle/fluid/tests/unittests/dist_fleet_ctr.py
@@ -62,26 +62,23 @@ class TestDistCTR2x2(FleetDistRunnerBase):
         """
         dnn_input_dim, lr_input_dim = int(1e5), int(1e5)
 
-        dnn_data = fluid.layers.data(
+        dnn_data = paddle.static.data(
             name="dnn_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        lr_data = fluid.layers.data(
+        lr_data = paddle.static.data(
             name="lr_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        label = fluid.layers.data(
+        label = paddle.static.data(
             name="click",
             shape=[-1, 1],
             dtype="int64",
             lod_level=0,
-            append_batch_size=False,
         )
 
         datas = [dnn_data, lr_data, label]
diff --git a/python/paddle/fluid/tests/unittests/dist_fleet_heter_pipeline_ctr.py b/python/paddle/fluid/tests/unittests/dist_fleet_heter_pipeline_ctr.py
index 8d4efa8c3d11629a49c034b46dab1fb21f9071cd..3e71a1cb6054d6a9a70f6e0c26e7ca4da67a4711 100644
--- a/python/paddle/fluid/tests/unittests/dist_fleet_heter_pipeline_ctr.py
+++ b/python/paddle/fluid/tests/unittests/dist_fleet_heter_pipeline_ctr.py
@@ -49,26 +49,23 @@ class TestHeterPipelinePsCTR2x2(FleetDistHeterRunnerBase):
         dnn_input_dim, lr_input_dim = int(1e5), int(1e5)
 
         with fluid.device_guard("cpu"):
-            dnn_data = fluid.layers.data(
+            dnn_data = paddle.static.data(
                 name="dnn_data",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
-            lr_data = fluid.layers.data(
+            lr_data = paddle.static.data(
                 name="lr_data",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
-            label = fluid.layers.data(
+            label = paddle.static.data(
                 name="click",
                 shape=[-1, 1],
                 dtype="float32",
                 lod_level=0,
-                append_batch_size=False,
             )
 
             datas = [dnn_data, lr_data, label]
diff --git a/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer.py b/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer.py
index 539446b6741b3cff2e6359587671f69c42b50f4c..dc0a7022b34348d7ee6aa40e2c89598cdc7ea0a3 100644
--- a/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer.py
@@ -74,8 +74,10 @@ def cnn_model(data):
 class TestFleetMetaOptimizerPrecision(TestDistRunnerBase):
     def get_model(self, batch_size=2, single_device=False):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer_fuse_allreduce.py b/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer_fuse_allreduce.py
index efad598c725a44c2bf5edf4ab2eb5d4df99d50cd..ac1a4c632fd495d40e905d7ac295f30d8d8fd7e5 100644
--- a/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer_fuse_allreduce.py
+++ b/python/paddle/fluid/tests/unittests/dist_fleet_raw_program_optimizer_fuse_allreduce.py
@@ -74,8 +74,10 @@ def cnn_model(data):
 class TestFleetMetaOptimizerFuseAllReducePrecision(TestDistRunnerBase):
     def get_model(self, batch_size=2, single_device=False):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_fleet_simnet_bow.py b/python/paddle/fluid/tests/unittests/dist_fleet_simnet_bow.py
index 358b3b5e39a5f747ca9856d22addb229990af931..b673bfeae16e2563333ac9e77d9f5d087b2fb734 100644
--- a/python/paddle/fluid/tests/unittests/dist_fleet_simnet_bow.py
+++ b/python/paddle/fluid/tests/unittests/dist_fleet_simnet_bow.py
@@ -93,18 +93,18 @@ def train_network(
     is_pyreader=False,
 ):
     # query
-    q = fluid.layers.data(
-        name="query_ids", shape=[1], dtype="int64", lod_level=1
+    q = paddle.static.data(
+        name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
     )
     # label data
-    label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+    label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
     # pt
-    pt = fluid.layers.data(
-        name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+    pt = paddle.static.data(
+        name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
     )
     # nt
-    nt = fluid.layers.data(
-        name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+    nt = paddle.static.data(
+        name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
     )
 
     datas = [q, label, pt, nt]
diff --git a/python/paddle/fluid/tests/unittests/dist_fleet_sparse_embedding_ctr.py b/python/paddle/fluid/tests/unittests/dist_fleet_sparse_embedding_ctr.py
index 9e398e83b90c48b6be430b642963615d5f770fae..a9a2d7be0ba413c337b6e2cd545f5e8e36764102 100644
--- a/python/paddle/fluid/tests/unittests/dist_fleet_sparse_embedding_ctr.py
+++ b/python/paddle/fluid/tests/unittests/dist_fleet_sparse_embedding_ctr.py
@@ -52,26 +52,23 @@ class TestDistCTR2x2(FleetDistRunnerBase):
         """
         dnn_input_dim, lr_input_dim = 10, 10
 
-        dnn_data = fluid.layers.data(
+        dnn_data = paddle.static.data(
             name="dnn_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        lr_data = fluid.layers.data(
+        lr_data = paddle.static.data(
             name="lr_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        label = fluid.layers.data(
+        label = paddle.static.data(
             name="click",
             shape=[-1, 1],
             dtype="int64",
             lod_level=0,
-            append_batch_size=False,
         )
 
         datas = [dnn_data, lr_data, label]
diff --git a/python/paddle/fluid/tests/unittests/dist_mnist.py b/python/paddle/fluid/tests/unittests/dist_mnist.py
index 117e178dd9670becda56bcae455f9426de42a9ae..87eb22dceac1ccca81765b83f43abb29060d0229 100644
--- a/python/paddle/fluid/tests/unittests/dist_mnist.py
+++ b/python/paddle/fluid/tests/unittests/dist_mnist.py
@@ -73,8 +73,10 @@ def cnn_model(data):
 class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2, use_dgc=False, dist_strategy=None):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_mnist_batch_merge.py b/python/paddle/fluid/tests/unittests/dist_mnist_batch_merge.py
index cac46996edd4a35a98addcbffd8506f89b1c709b..e38d2176bc18d3a8bfb3dd41d8fd4b5e57a118db 100644
--- a/python/paddle/fluid/tests/unittests/dist_mnist_batch_merge.py
+++ b/python/paddle/fluid/tests/unittests/dist_mnist_batch_merge.py
@@ -38,8 +38,10 @@ def test_merge_reader(repeat_batch_size=8):
 class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_mnist_fp16_allreduce.py b/python/paddle/fluid/tests/unittests/dist_mnist_fp16_allreduce.py
index e4af13fa89b066e02a9f026f3d6a11223a5536c1..9aa662854274f9b473142fbc0f85c5baa78156c1 100644
--- a/python/paddle/fluid/tests/unittests/dist_mnist_fp16_allreduce.py
+++ b/python/paddle/fluid/tests/unittests/dist_mnist_fp16_allreduce.py
@@ -32,8 +32,10 @@ fluid.default_main_program().random_seed = 1
 class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_mnist_lars.py b/python/paddle/fluid/tests/unittests/dist_mnist_lars.py
index dc2fc10450bf6e6532e68926d7851f888ad4f1d6..b1f5e0f2e8f3499d87a3e3f0d8aa72efdecac616 100644
--- a/python/paddle/fluid/tests/unittests/dist_mnist_lars.py
+++ b/python/paddle/fluid/tests/unittests/dist_mnist_lars.py
@@ -29,8 +29,10 @@ fluid.default_main_program().random_seed = 1
 class TestDistMnist2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2):
         # Input data
-        images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        images = paddle.static.data(
+            name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = cnn_model(images)
diff --git a/python/paddle/fluid/tests/unittests/dist_se_resnext.py b/python/paddle/fluid/tests/unittests/dist_se_resnext.py
index 377ad3072277dfda398d255c6ba31089700a679c..db3318d67d88ab8f629f12e42c3734f54fa0f19b 100644
--- a/python/paddle/fluid/tests/unittests/dist_se_resnext.py
+++ b/python/paddle/fluid/tests/unittests/dist_se_resnext.py
@@ -209,10 +209,10 @@ class SE_ResNeXt:
 class DistSeResneXt2x2(TestDistRunnerBase):
     def get_model(self, batch_size=2, use_dgc=False):
         # Input data
-        image = fluid.layers.data(
-            name="data", shape=[3, 224, 224], dtype='float32'
+        image = paddle.static.data(
+            name="data", shape=[-1, 3, 224, 224], dtype='float32'
         )
-        label = fluid.layers.data(name="int64", shape=[1], dtype='int64')
+        label = paddle.static.data(name="int64", shape=[-1, 1], dtype='int64')
 
         # Train program
         model = SE_ResNeXt(layers=50)
diff --git a/python/paddle/fluid/tests/unittests/dist_sharding_save.py b/python/paddle/fluid/tests/unittests/dist_sharding_save.py
index 1c4f49093df3530cc93a973e223df02be2ff4d9b..c509b2506b08daaa69323e1b5dc846cfaca3c06c 100755
--- a/python/paddle/fluid/tests/unittests/dist_sharding_save.py
+++ b/python/paddle/fluid/tests/unittests/dist_sharding_save.py
@@ -38,12 +38,10 @@ def runtime_main():
     fleet.init(role)
     with fluid.program_guard(train_prog, startup_prog):
         with fluid.unique_name.guard():
-            input_x = paddle.fluid.layers.data(
-                name="x", shape=[32], dtype='float32'
-            )
-            input_y = paddle.fluid.layers.data(
-                name="y", shape=[1], dtype='int64'
+            input_x = paddle.static.data(
+                name="x", shape=[-1, 32], dtype='float32'
             )
+            input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
             fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
             fc_2 = paddle.static.nn.fc(x=fc_1, size=256, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/dist_text_classification.py b/python/paddle/fluid/tests/unittests/dist_text_classification.py
index de8630cf70b49ef441adac752ca374a5959b16f1..a287bd8a6c878400ac29e9b15f278450fa5132d3 100644
--- a/python/paddle/fluid/tests/unittests/dist_text_classification.py
+++ b/python/paddle/fluid/tests/unittests/dist_text_classification.py
@@ -95,8 +95,8 @@ def conv_net(
 
 
 def inference_network(dict_dim):
-    data = fluid.layers.data(
-        name="words", shape=[1], dtype="int64", lod_level=1
+    data = paddle.static.data(
+        name="words", shape=[-1, 1], dtype="int64", lod_level=1
     )
     out = conv_net(data, dict_dim)
     return out
@@ -125,10 +125,10 @@ class TestDistTextClassification2x2(TestDistRunnerBase):
         word_dict, dict_dim = get_worddict(vocab)
 
         # Input data
-        data = fluid.layers.data(
-            name="words", shape=[1], dtype="int64", lod_level=1
+        data = paddle.static.data(
+            name="words", shape=[-1, 1], dtype="int64", lod_level=1
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         # Train program
         predict = conv_net(data, dict_dim)
diff --git a/python/paddle/fluid/tests/unittests/dist_transformer.py b/python/paddle/fluid/tests/unittests/dist_transformer.py
index 0db96bee8bba67ed0e63d32a5028c1681dbfdcf5..5b0343bd81c245d3780138146882973ec259ad10 100644
--- a/python/paddle/fluid/tests/unittests/dist_transformer.py
+++ b/python/paddle/fluid/tests/unittests/dist_transformer.py
@@ -1512,14 +1512,13 @@ def make_all_inputs(input_fields):
     """
     inputs = []
     for input_field in input_fields:
-        input_var = layers.data(
+        input_var = paddle.static.data(
             name=input_field,
             shape=input_descs[input_field][0],
             dtype=input_descs[input_field][1],
             lod_level=input_descs[input_field][2]
             if len(input_descs[input_field]) == 3
             else 0,
-            append_batch_size=False,
         )
         inputs.append(input_var)
     return inputs
diff --git a/python/paddle/fluid/tests/unittests/dist_word2vec.py b/python/paddle/fluid/tests/unittests/dist_word2vec.py
index 746a1f07655ab00330bda06e229ff0b28c612a35..e10131667c745fd6da97f3fb1a742d60b7fcf3df 100644
--- a/python/paddle/fluid/tests/unittests/dist_word2vec.py
+++ b/python/paddle/fluid/tests/unittests/dist_word2vec.py
@@ -107,13 +107,21 @@ class TestDistWord2vec2x2(TestDistRunnerBase):
         word_dict = paddle.dataset.imikolov.build_dict()
         dict_size = len(word_dict)
 
-        first_word = fluid.layers.data(name='firstw', shape=[1], dtype='int64')
-        second_word = fluid.layers.data(
-            name='secondw', shape=[1], dtype='int64'
+        first_word = paddle.static.data(
+            name='firstw', shape=[-1, 1], dtype='int64'
+        )
+        second_word = paddle.static.data(
+            name='secondw', shape=[-1, 1], dtype='int64'
+        )
+        third_word = paddle.static.data(
+            name='thirdw', shape=[-1, 1], dtype='int64'
+        )
+        forth_word = paddle.static.data(
+            name='forthw', shape=[-1, 1], dtype='int64'
+        )
+        next_word = paddle.static.data(
+            name='nextw', shape=[-1, 1], dtype='int64'
         )
-        third_word = fluid.layers.data(name='thirdw', shape=[1], dtype='int64')
-        forth_word = fluid.layers.data(name='forthw', shape=[1], dtype='int64')
-        next_word = fluid.layers.data(name='nextw', shape=[1], dtype='int64')
         avg_cost, predict_word = __network__(
             [first_word, second_word, third_word, forth_word, next_word]
         )
diff --git a/python/paddle/fluid/tests/unittests/distribution/test_distribution_normal.py b/python/paddle/fluid/tests/unittests/distribution/test_distribution_normal.py
index 42c69af5e6bebd9e7c23061328545c27be08f8ad..bb3fb57161a8cdfa0713e1ca1df6a01f1f7e6594 100644
--- a/python/paddle/fluid/tests/unittests/distribution/test_distribution_normal.py
+++ b/python/paddle/fluid/tests/unittests/distribution/test_distribution_normal.py
@@ -24,7 +24,6 @@ from test_distribution import DistributionNumpy
 import paddle
 from paddle import fluid
 from paddle.distribution import Normal
-from paddle.fluid import layers
 
 np.random.seed(2022)
 
@@ -117,8 +116,8 @@ class NormalTest(unittest.TestCase):
         self.static_other_loc = self.other_loc_np
         self.static_other_scale = self.other_scale_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1], dtype='float32'
             )
 
     def compare_with_numpy(self, fetch_list, sample_shape=7, tolerance=1e-6):
@@ -237,8 +236,8 @@ class NormalTest3(NormalTest):
         self.static_other_loc = self.other_loc_np
         self.static_other_scale = self.other_scale_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -266,8 +265,8 @@ class NormalTest4(NormalTest):
         self.static_other_loc = self.other_loc_np
         self.static_other_scale = self.other_scale_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -302,8 +301,8 @@ class NormalTest5(NormalTest):
         self.static_other_loc = self.other_loc_np
         self.static_other_scale = self.other_scale_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float64'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float64'
             )
 
 
@@ -334,20 +333,20 @@ class NormalTest6(NormalTest):
 
     def init_static_data(self, batch_size, dims):
         with fluid.program_guard(self.test_program):
-            self.static_loc = layers.data(
-                name='loc', shape=[dims], dtype='float32'
+            self.static_loc = paddle.static.data(
+                name='loc', shape=[-1, dims], dtype='float32'
             )
-            self.static_scale = layers.data(
-                name='scale', shape=[dims], dtype='float32'
+            self.static_scale = paddle.static.data(
+                name='scale', shape=[-1, dims], dtype='float32'
             )
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
-            self.static_other_loc = layers.data(
-                name='other_loc', shape=[dims], dtype='float32'
+            self.static_other_loc = paddle.static.data(
+                name='other_loc', shape=[-1, dims], dtype='float32'
             )
-            self.static_other_scale = layers.data(
-                name='other_scale', shape=[dims], dtype='float32'
+            self.static_other_scale = paddle.static.data(
+                name='other_scale', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -382,20 +381,20 @@ class NormalTest7(NormalTest):
 
     def init_static_data(self, batch_size, dims):
         with fluid.program_guard(self.test_program):
-            self.static_loc = layers.data(
-                name='loc', shape=[dims], dtype='float64'
+            self.static_loc = paddle.static.data(
+                name='loc', shape=[-1, dims], dtype='float64'
             )
-            self.static_scale = layers.data(
-                name='scale', shape=[dims], dtype='float64'
+            self.static_scale = paddle.static.data(
+                name='scale', shape=[-1, dims], dtype='float64'
             )
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float64'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float64'
             )
-            self.static_other_loc = layers.data(
-                name='other_loc', shape=[dims], dtype='float64'
+            self.static_other_loc = paddle.static.data(
+                name='other_loc', shape=[-1, dims], dtype='float64'
             )
-            self.static_other_scale = layers.data(
-                name='other_scale', shape=[dims], dtype='float64'
+            self.static_other_scale = paddle.static.data(
+                name='other_scale', shape=[-1, dims], dtype='float64'
             )
 
 
@@ -430,20 +429,20 @@ class NormalTest8(NormalTest):
 
     def init_static_data(self, batch_size, dims):
         with fluid.program_guard(self.test_program):
-            self.static_loc = layers.data(
-                name='loc', shape=[dims], dtype='float64'
+            self.static_loc = paddle.static.data(
+                name='loc', shape=[-1, dims], dtype='float64'
             )
-            self.static_scale = layers.data(
-                name='scale', shape=[dims], dtype='float64'
+            self.static_scale = paddle.static.data(
+                name='scale', shape=[-1, dims], dtype='float64'
             )
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
-            self.static_other_loc = layers.data(
-                name='other_loc', shape=[dims], dtype='float64'
+            self.static_other_loc = paddle.static.data(
+                name='other_loc', shape=[-1, dims], dtype='float64'
             )
-            self.static_other_scale = layers.data(
-                name='other_scale', shape=[dims], dtype='float64'
+            self.static_other_scale = paddle.static.data(
+                name='other_scale', shape=[-1, dims], dtype='float64'
             )
 
 
@@ -477,8 +476,8 @@ class NormalTest9(NormalTest):
         self.static_other_loc = self.other_loc_np
         self.static_other_scale = self.other_scale_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -512,8 +511,8 @@ class NormalTest10(NormalTest):
         self.static_other_loc = self.other_loc_np
         self.static_other_scale = self.other_scale_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
diff --git a/python/paddle/fluid/tests/unittests/distribution/test_distribution_uniform.py b/python/paddle/fluid/tests/unittests/distribution/test_distribution_uniform.py
index da5500ccefb0edc65a4df7d517e35d1a932ae8b7..ca59b5118e67754ec71e020b48ea1ce601b54714 100644
--- a/python/paddle/fluid/tests/unittests/distribution/test_distribution_uniform.py
+++ b/python/paddle/fluid/tests/unittests/distribution/test_distribution_uniform.py
@@ -20,7 +20,6 @@ from test_distribution import DistributionNumpy
 import paddle
 from paddle import fluid
 from paddle.distribution import Uniform
-from paddle.fluid import layers
 
 np.random.seed(2022)
 
@@ -88,8 +87,8 @@ class UniformTest(unittest.TestCase):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1], dtype='float32'
             )
 
     def compare_with_numpy(self, fetch_list, sample_shape=7, tolerance=1e-6):
@@ -170,8 +169,8 @@ class UniformTest3(UniformTest):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -188,8 +187,8 @@ class UniformTest4(UniformTest):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -211,8 +210,8 @@ class UniformTest5(UniformTest):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float64'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float64'
             )
 
 
@@ -232,14 +231,14 @@ class UniformTest6(UniformTest):
 
     def init_static_data(self, batch_size, dims):
         with fluid.program_guard(self.test_program):
-            self.static_low = layers.data(
-                name='low', shape=[dims], dtype='float32'
+            self.static_low = paddle.static.data(
+                name='low', shape=[-1, dims], dtype='float32'
             )
-            self.static_high = layers.data(
-                name='high', shape=[dims], dtype='float32'
+            self.static_high = paddle.static.data(
+                name='high', shape=[-1, dims], dtype='float32'
             )
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -259,14 +258,14 @@ class UniformTest7(UniformTest):
 
     def init_static_data(self, batch_size, dims):
         with fluid.program_guard(self.test_program):
-            self.static_low = layers.data(
-                name='low', shape=[dims], dtype='float64'
+            self.static_low = paddle.static.data(
+                name='low', shape=[-1, dims], dtype='float64'
             )
-            self.static_high = layers.data(
-                name='high', shape=[dims], dtype='float64'
+            self.static_high = paddle.static.data(
+                name='high', shape=[-1, dims], dtype='float64'
             )
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float64'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float64'
             )
 
 
@@ -286,14 +285,14 @@ class UniformTest8(UniformTest):
 
     def init_static_data(self, batch_size, dims):
         with fluid.program_guard(self.test_program):
-            self.static_low = layers.data(
-                name='low', shape=[dims], dtype='float64'
+            self.static_low = paddle.static.data(
+                name='low', shape=[-1, dims], dtype='float64'
             )
-            self.static_high = layers.data(
-                name='high', shape=[dims], dtype='float64'
+            self.static_high = paddle.static.data(
+                name='high', shape=[-1, dims], dtype='float64'
             )
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -311,8 +310,8 @@ class UniformTest9(UniformTest):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -333,8 +332,8 @@ class UniformTest10(UniformTest):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
@@ -355,8 +354,8 @@ class UniformTest11(UniformTest):
         self.static_low = self.low_np
         self.static_high = self.high_np
         with fluid.program_guard(self.test_program):
-            self.static_values = layers.data(
-                name='values', shape=[dims], dtype='float32'
+            self.static_values = paddle.static.data(
+                name='values', shape=[-1, dims], dtype='float32'
             )
 
 
diff --git a/python/paddle/fluid/tests/unittests/dygraph_to_static/transformer_util.py b/python/paddle/fluid/tests/unittests/dygraph_to_static/transformer_util.py
index 92bdde2d94f9eaf29a8ed5cb16c85b62f419389c..680d8afd4860ff61e924722e4bc73a303bb42f65 100644
--- a/python/paddle/fluid/tests/unittests/dygraph_to_static/transformer_util.py
+++ b/python/paddle/fluid/tests/unittests/dygraph_to_static/transformer_util.py
@@ -296,12 +296,11 @@ class InputField:
         self.feed_list = []
         for slot in input_slots:
             self.feed_list.append(
-                fluid.layers.data(
+                paddle.static.data(
                     name=slot['name'],
                     shape=slot['shape'],
                     dtype=slot['dtype'],
                     lod_level=slot.get('lod_level', 0),
-                    append_batch_size=False,
                 )
             )
 
diff --git a/python/paddle/fluid/tests/unittests/fleet_heter_ps_training.py b/python/paddle/fluid/tests/unittests/fleet_heter_ps_training.py
index 8f803ce1db1424826fdbed39f67b158d1f6d56db..917beec752d2aa080614d04d8aeadd010a90308b 100644
--- a/python/paddle/fluid/tests/unittests/fleet_heter_ps_training.py
+++ b/python/paddle/fluid/tests/unittests/fleet_heter_ps_training.py
@@ -41,26 +41,23 @@ def net(batch_size=4, lr=0.01):
     dnn_input_dim, lr_input_dim = int(2), int(2)
 
     with fluid.device_guard("cpu"):
-        dnn_data = fluid.layers.data(
+        dnn_data = paddle.static.data(
             name="dnn_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        lr_data = fluid.layers.data(
+        lr_data = paddle.static.data(
             name="lr_data",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
-        label = fluid.layers.data(
+        label = paddle.static.data(
             name="click",
             shape=[-1, 1],
             dtype="float32",
             lod_level=0,
-            append_batch_size=False,
         )
 
         datas = [dnn_data, lr_data, label]
diff --git a/python/paddle/fluid/tests/unittests/fleet_meta_optimizer_base.py b/python/paddle/fluid/tests/unittests/fleet_meta_optimizer_base.py
index 628953391c059fd4b4b5652acc8cbdaa79213ffe..e17a9604fc5c6f9951297d4924d6a87dc5b6d7d6 100755
--- a/python/paddle/fluid/tests/unittests/fleet_meta_optimizer_base.py
+++ b/python/paddle/fluid/tests/unittests/fleet_meta_optimizer_base.py
@@ -55,11 +55,11 @@ class TestFleetMetaOptimizer(unittest.TestCase):
             with fluid.unique_name.guard():
                 role = role_maker.PaddleCloudRoleMaker(is_collective=True)
                 fleet.init(role)
-                input_x = paddle.fluid.layers.data(
-                    name="x", shape=[32], dtype='float32'
+                input_x = paddle.static.data(
+                    name="x", shape=[-1, 32], dtype='float32'
                 )
-                input_y = paddle.fluid.layers.data(
-                    name="y", shape=[1], dtype='int64'
+                input_y = paddle.static.data(
+                    name="y", shape=[-1, 1], dtype='int64'
                 )
 
                 fc_1 = paddle.static.nn.fc(
@@ -92,11 +92,11 @@ class TestFleetMetaOptimizer(unittest.TestCase):
                 role = role_maker.PaddleCloudRoleMaker(is_collective=True)
                 fleet.init(role)
                 with fluid.device_guard("gpu:0"):
-                    input_x = paddle.fluid.layers.data(
-                        name="x", shape=[32], dtype='float32'
+                    input_x = paddle.static.data(
+                        name="x", shape=[-1, 32], dtype='float32'
                     )
-                    input_y = paddle.fluid.layers.data(
-                        name="y", shape=[1], dtype='int64'
+                    input_y = paddle.static.data(
+                        name="y", shape=[-1, 1], dtype='int64'
                     )
 
                 for stage_idx in range(pp_degree):
diff --git a/python/paddle/fluid/tests/unittests/fleet_ps_training.py b/python/paddle/fluid/tests/unittests/fleet_ps_training.py
index be097ad3ac530f36f944d89cd13d62f846743187..9b5ccf7c99f3af6807a9a50b364d5b66321e03da 100644
--- a/python/paddle/fluid/tests/unittests/fleet_ps_training.py
+++ b/python/paddle/fluid/tests/unittests/fleet_ps_training.py
@@ -15,14 +15,15 @@
 from nets import mlp
 from utils import gen_data
 
+import paddle
 import paddle.fluid as fluid
 from paddle.fluid.incubate.fleet.base import role_maker
 from paddle.fluid.incubate.fleet.parameter_server.distribute_transpiler import (
     fleet,
 )
 
-input_x = fluid.layers.data(name="x", shape=[32], dtype='float32')
-input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 input_y = fluid.layers.cast(input_y, dtype="float32")
 
 with fluid.device_guard("gpu"):
diff --git a/python/paddle/fluid/tests/unittests/ir/test_ir_embedding_eltwise_layernorm_fuse_pass.py b/python/paddle/fluid/tests/unittests/ir/test_ir_embedding_eltwise_layernorm_fuse_pass.py
index ba5e304b61155765523c418eac2157fd77b90108..8f6c8592423047e40abcc4258a51aaad43c95eb5 100644
--- a/python/paddle/fluid/tests/unittests/ir/test_ir_embedding_eltwise_layernorm_fuse_pass.py
+++ b/python/paddle/fluid/tests/unittests/ir/test_ir_embedding_eltwise_layernorm_fuse_pass.py
@@ -25,23 +25,20 @@ import paddle.fluid.core as core
 class EmbEltwiseLayerNormFusePassTest(PassTest):
     def setUp(self):
         with fluid.program_guard(self.main_program, self.startup_program):
-            word_id = fluid.layers.data(
+            word_id = paddle.static.data(
                 name="word_id",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
-            pos_id = fluid.layers.data(
+            pos_id = paddle.static.data(
                 name="pos_id",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
-            sent_id = fluid.layers.data(
+            sent_id = paddle.static.data(
                 name="sent_id",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
             word_emb = fluid.layers.embedding(
                 input=word_id, size=(128, 768), dtype='float32'
@@ -56,29 +53,25 @@ class EmbEltwiseLayerNormFusePassTest(PassTest):
             add2 = paddle.add(add1, sent_emb)
             hidden1 = paddle.static.nn.layer_norm(input=add2, begin_norm_axis=2)
 
-            id1 = fluid.layers.data(
+            id1 = paddle.static.data(
                 name="id1",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
-            id2 = fluid.layers.data(
+            id2 = paddle.static.data(
                 name="id2",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
-            id3 = fluid.layers.data(
+            id3 = paddle.static.data(
                 name="id3",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
-            id4 = fluid.layers.data(
+            id4 = paddle.static.data(
                 name="id4",
                 shape=[1, 128, 1],
                 dtype="int64",
-                append_batch_size=False,
             )
             emb1 = fluid.layers.embedding(
                 input=id1, size=(128, 768), dtype='float32'
diff --git a/python/paddle/fluid/tests/unittests/ir/test_ir_subgraph_python_interface.py b/python/paddle/fluid/tests/unittests/ir/test_ir_subgraph_python_interface.py
index 21697177d0dd626f9f81c555e8dca2dce4bfe3a2..6e0bb08c926b3180d64812e66db403b8751c4bbb 100644
--- a/python/paddle/fluid/tests/unittests/ir/test_ir_subgraph_python_interface.py
+++ b/python/paddle/fluid/tests/unittests/ir/test_ir_subgraph_python_interface.py
@@ -28,10 +28,12 @@ paddle.enable_static()
 class TestQuantizationSubGraph(unittest.TestCase):
     def build_graph_with_sub_graph(self):
         def linear_fc(num):
-            data = fluid.layers.data(
-                name='image', shape=[1, 32, 32], dtype='float32'
+            data = paddle.static.data(
+                name='image', shape=[-1, 1, 32, 32], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             hidden = data
             for _ in range(num):
                 hidden = paddle.static.nn.fc(
diff --git a/python/paddle/fluid/tests/unittests/ir_memory_optimize_net_base.py b/python/paddle/fluid/tests/unittests/ir_memory_optimize_net_base.py
index 9f7b2d8533573e8f68afe9e49477790e0c80e6de..ae2870af8315b06c56854a7f53d160c9216d9f15 100644
--- a/python/paddle/fluid/tests/unittests/ir_memory_optimize_net_base.py
+++ b/python/paddle/fluid/tests/unittests/ir_memory_optimize_net_base.py
@@ -60,11 +60,11 @@ class BuildIrMemOptBase(unittest.TestCase):
         fluid.default_startup_program().random_seed = 100
         fluid.default_main_program().random_seed = 100
 
-        data = fluid.layers.data(
-            name="words", shape=[1], dtype="int64", lod_level=1
+        data = paddle.static.data(
+            name="words", shape=[-1, 1], dtype="int64", lod_level=1
         )
 
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
 
         cost = network(data, label, len(self.word_dict))
         optimizer = fluid.optimizer.Adam(learning_rate=0.001)
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_allgather_api.py b/python/paddle/fluid/tests/unittests/mlu/collective_allgather_api.py
index 1d8d11ad1120008c67ef497b698608a4c6b9233e..91c839e9ea5310deef48a73571bcd2ec4df79e0b 100755
--- a/python/paddle/fluid/tests/unittests/mlu/collective_allgather_api.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_allgather_api.py
@@ -45,8 +45,8 @@ class TestCollectiveAllgatherAPI(TestCollectiveAPIRunnerBase):
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
             tensor_list = []
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             paddle.distributed.all_gather(tensor_list, tindata)
             return tensor_list
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_allgather_op.py b/python/paddle/fluid/tests/unittests/mlu/collective_allgather_op.py
index 8040e834d65a90b7f6e883fff481008963f084f7..33d9f05cf7b2fa6603f99c6d974a2aeb88ebdf83 100755
--- a/python/paddle/fluid/tests/unittests/mlu/collective_allgather_op.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_allgather_op.py
@@ -42,9 +42,10 @@ class TestCollectiveAllgather(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
             toutdata = main_prog.current_block().create_var(
                 name="outofallgather",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_api.py b/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_api.py
index 22ca990c55afde93e2b28ab1167785e6e0559f49..a636dbb4692f6d721c83cbfd20bc60376b6cdede 100644
--- a/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_api.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_api.py
@@ -44,8 +44,8 @@ class TestCollectiveAllreduceAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             paddle.distributed.all_reduce(tindata)
             return [tindata]
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_op.py b/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_op.py
index 609d9e7c41688e40a16b8216a8a69bc9c4b7815c..2d9f6ee03e0a836e18b626aaf2ab1ecc7eafe521 100644
--- a/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_op.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_allreduce_op.py
@@ -42,9 +42,11 @@ class TestCollectiveAllreduce(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program, col_type):
         ring_id = 0
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
+            
             toutdata = main_prog.current_block().create_var(
                 name="outof" + col_type,
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_api.py b/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_api.py
index 0c1ae572251aa6dc579090ead3433b73178f79c5..673ffe67a0f7b9a3744e48b85e36db8cafa46e87 100644
--- a/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_api.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_api.py
@@ -44,9 +44,11 @@ class TestCollectiveBroadcastAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype="float32"
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype="float32"
             )
+            tindata.desc.set_need_check_feed(False)
+            
             paddle.distributed.broadcast(tindata, src=1)
             return [tindata]
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_op.py b/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_op.py
index 982d5d204fccc23d0a66bbb605135270c51534e4..ed42a7964ea1a8589efb148a360eb32d4826bac8 100755
--- a/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_op.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_broadcast_op.py
@@ -43,9 +43,11 @@ class TestCollectiveBroadcast(TestCollectiveRunnerBase):
         ring_id = 0
         rootid = 1
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
+            
             toutdata = main_prog.current_block().create_var(
                 name="outofbroadcast",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_reduce_api.py b/python/paddle/fluid/tests/unittests/mlu/collective_reduce_api.py
index 33a0d9d0d51a7efb0724c7d04c80e32a2f3c0480..2121aaee2bcd8da9113cab65dd602c2ef868cc1c 100644
--- a/python/paddle/fluid/tests/unittests/mlu/collective_reduce_api.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_reduce_api.py
@@ -44,9 +44,11 @@ class TestCollectiveReduceAPI(TestCollectiveAPIRunnerBase):
 
     def get_model(self, main_prog, startup_program, rank):
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
+            
             paddle.distributed.reduce(tindata, dst=0)
             return [tindata]
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/collective_reduce_op.py b/python/paddle/fluid/tests/unittests/mlu/collective_reduce_op.py
index d9271376b0f3124427003d4ef90cc7a3ae87404a..f389cef9d6f8884ef0b43139b93bc9f96d72d2fa 100644
--- a/python/paddle/fluid/tests/unittests/mlu/collective_reduce_op.py
+++ b/python/paddle/fluid/tests/unittests/mlu/collective_reduce_op.py
@@ -43,9 +43,11 @@ class TestCollectiveReduce(TestCollectiveRunnerBase):
         ring_id = 0
         rootid = 1
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
+            tindata.desc.set_need_check_feed(False)
+            
             toutdata = main_prog.current_block().create_var(
                 name="outof" + col_type,
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/mlu/sync_batch_norm_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/sync_batch_norm_op_mlu.py
index 20b3a00cc947b3139f236cf799d45f0e83b0efa8..8a9405c1fcb3e007214e04dce175519452188d35 100644
--- a/python/paddle/fluid/tests/unittests/mlu/sync_batch_norm_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/sync_batch_norm_op_mlu.py
@@ -70,11 +70,10 @@ class TestSyncBatchNormOpTraining(TestSyncBatchNormRunnerBase):
         use_cudnn = False
         with fluid.unique_name.guard():
             with fluid.program_guard(main, startup):
-                data = fluid.layers.data(
+                data = paddle.static.data(
                     name='input',
                     shape=self.dshape,
                     dtype=self.dtype,
-                    append_batch_size=False,
                 )
                 conv = paddle.static.nn.conv2d(
                     input=data,
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_accuracy_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_accuracy_op_mlu.py
index a32765b8d306894bdfe843330337ae5536f8ed5b..644a2e81dbffa4d1e2bf6a4c9648bb7951b8c57f 100755
--- a/python/paddle/fluid/tests/unittests/mlu/test_accuracy_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_accuracy_op_mlu.py
@@ -71,16 +71,16 @@ class TestAccuracyOpError(unittest.TestCase):
             x1 = fluid.create_lod_tensor(
                 np.array([[-1]]), [[1]], fluid.MLUPlace(0)
             )
-            label = fluid.layers.data(
+            label = paddle.static.data(
                 name='label', shape=[-1, 1], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.static.accuracy, x1, label)
             self.assertRaises(TypeError, paddle.metric.accuracy, x1, label)
             # The input dtype of accuracy_op must be float32 or float64.
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="int32")
             self.assertRaises(TypeError, paddle.static.accuracy, x2, label)
             self.assertRaises(TypeError, paddle.metric.accuracy, x2, label)
-            x3 = fluid.layers.data(name='input', shape=[-1, 2], dtype="float16")
+            x3 = paddle.static.data(name='input', shape=[-1, 2], dtype="float16")
             paddle.static.accuracy(input=x3, label=label)
             paddle.metric.accuracy(input=x3, label=label)
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_batch_norm_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_batch_norm_op_mlu.py
index 66876ddb7929452ddac4225e4095cc11fdd6442c..abd86efcf84d08017172a5af16844b2985c2e379 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_batch_norm_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_batch_norm_op_mlu.py
@@ -745,7 +745,7 @@ class TestBatchNormOpError(unittest.TestCase):
 
             # the input dtype of batch_norm must be float16 or float32 or float64
             # float16 only can be set on GPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32")
             self.assertRaises(TypeError, paddle.static.nn.batch_norm, x2)
 
 
@@ -761,7 +761,7 @@ class TestDygraphBatchNormAPIError(unittest.TestCase):
 
             # the input dtype of BatchNorm must be float16 or float32 or float64
             # float16 only can be set on GPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32")
             self.assertRaises(TypeError, batch_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_compare_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_compare_op_mlu.py
index f994fb185f64efe22d34ebb75688e6a1c2408d70..568d03acd4e638aa7c621b7efed622e3b668a1b0 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_compare_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_compare_op_mlu.py
@@ -44,9 +44,9 @@ def create_test_class(op_type, typename, callback):
         def test_errors(self):
             paddle.enable_static()
             with program_guard(Program(), Program()):
-                a = fluid.layers.data(name='a', shape=[2], dtype='float32')
-                b = fluid.layers.data(name='b', shape=[2], dtype='float32')
-                c = fluid.layers.data(name='c', shape=[2], dtype='int16')
+                a = paddle.static.data(name='a', shape=[-1, 2], dtype='float32')
+                b = paddle.static.data(name='b', shape=[-1, 2], dtype='float32')
+                c = paddle.static.data(name='c', shape=[-1, 2], dtype='int16')
                 d = fluid.create_lod_tensor(np.array([[-1]]), [[1]], self.place)
 
                 op = eval("fluid.layers.%s" % self.op_type)
@@ -134,8 +134,8 @@ def create_test_class(op_type, typename, callback):
         def test_attr_name(self):
             paddle.enable_static()
             with program_guard(Program(), Program()):
-                x = fluid.layers.data(name='x', shape=[4], dtype=typename)
-                y = fluid.layers.data(name='y', shape=[4], dtype=typename)
+                x = paddle.static.data(name='x', shape=[-1, 4], dtype=typename)
+                y = paddle.static.data(name='y', shape=[-1, 4], dtype=typename)
                 op = eval("paddle.%s" % (self.op_type))
                 out = op(x=x, y=y, name="name_%s" % (self.op_type))
             self.assertEqual("name_%s" % (self.op_type) in out.name, True)
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_conv2d_transposed_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_conv2d_transposed_op_mlu.py
index a24206e0900b37e949ba6a018c30ae36b9d86d70..b4329068c62e0a3ffdc3805377fa085737cf0716 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_conv2d_transposed_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_conv2d_transposed_op_mlu.py
@@ -493,11 +493,11 @@ class TestConv2DTransposeAPI(unittest.TestCase):
         self.place = paddle.device.MLUPlace(0)
 
     def test_case1(self):
-        data1 = fluid.layers.data(
-            name='data1', shape=[3, 5, 5], dtype='float32'
+        data1 = paddle.static.data(
+            name='data1', shape=[-1, 3, 5, 5], dtype='float32'
         )
-        data2 = fluid.layers.data(
-            name='data2', shape=[5, 5, 3], dtype='float32'
+        data2 = paddle.static.data(
+            name='data2', shape=[-1, 5, 5, 3], dtype='float32'
         )
         out1 = paddle.static.nn.conv2d_transpose(
             input=data1,
@@ -583,7 +583,7 @@ class TestConv2DTransposeOpException(unittest.TestCase):
         self.place = paddle.device.MLUPlace(0)
 
     def test_exception(self):
-        data = fluid.layers.data(name='data', shape=[3, 5, 5], dtype="float32")
+        data = paddle.static.data(name='data', shape=[-1, 3, 5, 5], dtype="float32")
 
         def attr_data_format():
             out = paddle.static.nn.conv2d_transpose(
@@ -630,8 +630,8 @@ class TestConv2DTransposeOpException(unittest.TestCase):
 
         self.assertRaises(ValueError, attr_padding_with_data_format)
 
-        error_input = fluid.layers.data(
-            name='error_data', shape=[1], dtype="float32"
+        error_input = paddle.static.data(
+            name='error_data', shape=[-1, 1], dtype="float32"
         )
 
         def error_input_size():
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_expand_as_v2_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_expand_as_v2_op_mlu.py
index 0a88f759831e4672f7afec6d4a12796f97c2311f..5d7b39925665c20675f028b42a4a0347e2825dfe 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_expand_as_v2_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_expand_as_v2_op_mlu.py
@@ -137,14 +137,13 @@ class TestExpandAsV2API(unittest.TestCase):
     def test_api(self):
         input1 = np.random.random([12, 14]).astype("float32")
         input2 = np.random.random([2, 12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
+        x = paddle.static.data(
+            name='x', shape=[12, 14], dtype="float32"
         )
 
-        y = fluid.layers.data(
+        y = paddle.static.data(
             name='target_tensor',
             shape=[2, 12, 14],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_expand_v2_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_expand_v2_op_mlu.py
index 46d3d9d214fd7d05a021892692ad7189a1a3690f..c2a3dbfa20944c75a671f9b6eda43684a95e68ee 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_expand_v2_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_expand_v2_op_mlu.py
@@ -220,9 +220,9 @@ class TestExpandV2Error(unittest.TestCase):
             )
             shape = [2, 2]
             self.assertRaises(TypeError, paddle.tensor.expand, x1, shape)
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="uint8")
             self.assertRaises(TypeError, paddle.tensor.expand, x2, shape)
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tensor.expand, x3, shape)
 
@@ -231,15 +231,14 @@ class TestExpandV2Error(unittest.TestCase):
 class TestExpandV2API(unittest.TestCase):
     def test_api(self):
         input = np.random.random([12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
+        x = paddle.static.data(
+            name='x', shape=[12, 14], dtype="float32"
         )
 
         positive_2 = fluid.layers.fill_constant([1], "int32", 12)
-        expand_shape = fluid.layers.data(
+        expand_shape = paddle.static.data(
             name="expand_shape",
             shape=[2],
-            append_batch_size=False,
             dtype="int32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_fill_constant_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_fill_constant_op_mlu.py
index 7a19779b049ce4d8874f1e4559b67b7149ba3cb4..6d67b6fc8f095a4c7387bad97809cadfb24e1b24 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_fill_constant_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_fill_constant_op_mlu.py
@@ -391,7 +391,7 @@ class TestFillConstantOpError(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
             # for ci coverage
-            x1 = fluid.layers.data(name='x1', shape=[1], dtype="int16")
+            x1 = paddle.static.data(name='x1', shape=[-1, 1], dtype="int16")
             self.assertRaises(
                 TypeError,
                 fluid.layers.fill_constant,
@@ -411,7 +411,7 @@ class TestFillConstantOpError(unittest.TestCase):
 
             # The argument dtype of fill_constant_op must be one of bool, float16,
             # float32, float64, uint8, int16, int32 or int64
-            x2 = fluid.layers.data(name='x2', shape=[1], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 1], dtype="int32")
 
             self.assertRaises(
                 TypeError,
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_layer_norm_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_layer_norm_op_mlu.py
index 6e095c9b85144c7b61f3bebf8b4269f6df80489b..e9d248928556b9eacddfb564ed4a4a8a55aeacdc 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_layer_norm_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_layer_norm_op_mlu.py
@@ -245,11 +245,10 @@ class TestLayerNormOp(unittest.TestCase):
 
 class TestLayerNormAPI(unittest.TestCase):
     def test_case(self):
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name='x',
             shape=[64, 32, 256],
             dtype='float32',
-            append_batch_size=False,
         )
         x = paddle.static.nn.layer_norm(
             x,
@@ -291,7 +290,7 @@ class TestDygraphLayerNormAPIError(unittest.TestCase):
             self.assertRaises(TypeError, layer_norm, x1)
 
             # the input dtype of LayerNorm must be float32 or float16
-            x2 = fluid.layers.data(name='x2', shape=[3, 32, 32], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 3, 32, 32], dtype="int32")
             self.assertRaises(TypeError, layer_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_log_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_log_op_mlu.py
index 11df0949a6f47151ee38c4efd8ff162be90b1e7d..c167ac67cc436102ad9f50f4953f9eb052a3a17e 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_log_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_log_op_mlu.py
@@ -70,11 +70,11 @@ class TestLog(TestActivation):
         self.outputs = {'Out': out}
 
     def test_error(self):
-        in1 = fluid.layers.data(
-            name="in1", shape=[11, 17], append_batch_size=False, dtype="int32"
+        in1 = paddle.static.data(
+            name="in1", shape=[11, 17], dtype="int32"
         )
-        in2 = fluid.layers.data(
-            name="in2", shape=[11, 17], append_batch_size=False, dtype="int64"
+        in2 = paddle.static.data(
+            name="in2", shape=[-1, 11, 17], dtype="int64"
         )
 
         self.assertRaises(TypeError, fluid.layers.log, in1)
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_momentum_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_momentum_op_mlu.py
index 5e5e517878eb3dbece3cc924e416f0cbce80418d..3a757e9cb66d86547095a874451e676eeb88f82f 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_momentum_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_momentum_op_mlu.py
@@ -140,8 +140,8 @@ class TestMomentumV2(unittest.TestCase):
         place = fluid.MLUPlace(0)
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1)
             cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
@@ -265,8 +265,8 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
         place = fluid.MLUPlace(0)
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1)
             cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_one_hot_v2_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_one_hot_v2_op_mlu.py
index b28d191fa2c3cd6f686eced3f33039e1dad8bc71..dfe989fa83b7b4443a6d65b747c322e23864028a 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_one_hot_v2_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_one_hot_v2_op_mlu.py
@@ -150,8 +150,8 @@ class TestOneHotOp_exception(unittest.TestCase):
     def test_check_output(self):
         program = Program()
         with program_guard(program):
-            x = fluid.layers.data(
-                name='x', shape=[self.dimension], dtype='float32', lod_level=1
+            x = paddle.static.data(
+                name='x', shape=[-1, self.dimension], dtype='float32', lod_level=1
             )
             block = program.current_block()
             one_hot_out = block.create_var(
@@ -207,7 +207,7 @@ class TestOneHotOpApi(unittest.TestCase):
             #     paddle.to_tensor(label), depth)
 
     def _run(self, depth):
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         one_hot_label = fluid.one_hot(input=label, depth=depth)
 
         label_data = np.array(
@@ -234,10 +234,9 @@ class BadInputTestOnehotV2(unittest.TestCase):
         with fluid.program_guard(fluid.Program()):
 
             def test_bad_x():
-                label = fluid.layers.data(
+                label = paddle.static.data(
                     name="label",
                     shape=[4],
-                    append_batch_size=False,
                     dtype="float32",
                 )
                 one_hot_label = fluid.one_hot(input=label, depth=4)
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_pool2d_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_pool2d_op_mlu.py
index 139d347e0d5da80d43c041314d865f6a480aad7c..4e4f0c56615d290b4b9103b7408f17348afb1637 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_pool2d_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_pool2d_op_mlu.py
@@ -705,31 +705,27 @@ class TestPool2DAPI(unittest.TestCase):
         x_NHWC = np.random.random([2, 5, 5, 3]).astype("float32")
         x_NCHW = np.random.random([2, 3, 5, 5]).astype("float32")
 
-        input_NHWC = fluid.layers.data(
+        input_NHWC = paddle.static.data(
             name="input_NHWC",
             shape=[2, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NCHW = fluid.layers.data(
+        input_NCHW = paddle.static.data(
             name="input_NCHW",
             shape=[2, 3, 5, 5],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NHWC_negetive = fluid.layers.data(
+        input_NHWC_negetive = paddle.static.data(
             name="input_NHWC_negetive",
             shape=[2, -1, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NCHW_negetive = fluid.layers.data(
+        input_NCHW_negetive = paddle.static.data(
             name="input_NCHW_negetive",
             shape=[2, 3, -1, -1],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -930,10 +926,9 @@ class TestPool2DAPI(unittest.TestCase):
 
 class TestPool2DAPI_Error(unittest.TestCase):
     def test_api(self):
-        input_NHWC = fluid.layers.data(
+        input_NHWC = paddle.static.data(
             name="input_NHWC",
             shape=[2, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
         ksize = [3, 3]
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_split_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_split_op_mlu.py
index 072b985613e0818075fa1d51583939665b0434a4..83d18b667385307ac36630a4f09e82e60de5b0c9 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_split_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_split_op_mlu.py
@@ -94,7 +94,7 @@ class TestCase5(TestCase1):
 class API_TestSplit(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float32')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float32')
             x0, x1 = paddle.split(data, num_or_sections=(3, 7), axis=1)
             place = fluid.MLUPlace(0)
             exe = fluid.Executor(place)
@@ -108,7 +108,7 @@ class API_TestSplit(unittest.TestCase):
 class API_TestSplit2(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float32')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float32')
             x0, x1 = paddle.split(data, num_or_sections=2, axis=1)
             place = fluid.MLUPlace(0)
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_stack_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_stack_op_mlu.py
index eefe1d7d691fcdc2325637d38349a50dc58f6c08..8bbd9e77897d9a505c668c2ce9b79c27dd784368 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_stack_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_stack_op_mlu.py
@@ -120,9 +120,9 @@ class TestStackOpHalf(TestStackOpBase):
 class API_test(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[1, 2], dtype='float32')
-            data2 = fluid.layers.data('data2', shape=[1, 2], dtype='float32')
-            data3 = fluid.layers.data('data3', shape=[1, 2], dtype='float32')
+            data1 = paddle.static.data('data1', shape=[-1, 1, 2], dtype='float32')
+            data2 = paddle.static.data('data2', shape=[-1, 1, 2], dtype='float32')
+            data3 = paddle.static.data('data3', shape=[-1, 1, 2], dtype='float32')
             result_stack = paddle.stack([data1, data2, data3], axis=0)
             place = paddle.MLUPlace(0)
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_strided_slice_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_strided_slice_op_mlu.py
index 35069485054f58079924aee28321397dc17e8f51..07a9179c5a8b112d94eefde4e454f5ac6919a089 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_strided_slice_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_strided_slice_op_mlu.py
@@ -527,20 +527,19 @@ class TestStridedSliceAPI(unittest.TestCase):
         input = np.random.random([3, 4, 5, 6]).astype("float32")
         minus_1 = fluid.layers.fill_constant([1], "int32", -1)
         minus_3 = fluid.layers.fill_constant([1], "int32", -3)
-        starts = fluid.layers.data(
-            name='starts', shape=[3], dtype='int32', append_batch_size=False
+        starts = paddle.static.data(
+            name='starts', shape=[3], dtype='int32'
         )
-        ends = fluid.layers.data(
-            name='ends', shape=[3], dtype='int32', append_batch_size=False
+        ends = paddle.static.data(
+            name='ends', shape=[3], dtype='int32'
         )
-        strides = fluid.layers.data(
-            name='strides', shape=[3], dtype='int32', append_batch_size=False
+        strides = paddle.static.data(
+            name='strides', shape=[3], dtype='int32'
         )
 
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[3, 4, 5, 6],
-            append_batch_size=False,
             dtype="float32",
         )
         out_1 = paddle.strided_slice(
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_sync_batch_norm_op_mlu_extra.py b/python/paddle/fluid/tests/unittests/mlu/test_sync_batch_norm_op_mlu_extra.py
index d9ec53af9b489d70614a8595891c16f02b92ca28..eb048e1ee60588e9b60957fddf553c3f6b648919 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_sync_batch_norm_op_mlu_extra.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_sync_batch_norm_op_mlu_extra.py
@@ -46,7 +46,7 @@ class TestDygraphSyncBatchNormAPIError(unittest.TestCase):
             self.assertRaises(TypeError, my_sync_batch_norm, x1)
 
             # the input dtype of SyncBatchNorm must be float16 or float32
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32")
             self.assertRaises(TypeError, my_sync_batch_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_tile_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_tile_op_mlu.py
index 2f1a0d990bf01a262a72433d7d949c6d7e28d0ad..c3f4b8a755558f191d21aa869a7deb2bd165f104 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_tile_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_tile_op_mlu.py
@@ -221,9 +221,9 @@ class TestTileError(unittest.TestCase):
             )
             repeat_times = [2, 2]
             self.assertRaises(TypeError, paddle.tile, x1, repeat_times)
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="uint8")
             self.assertRaises(TypeError, paddle.tile, x2, repeat_times)
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tile, x3, repeat_times)
 
@@ -232,7 +232,7 @@ class TestTileAPIStatic(unittest.TestCase):
     def test_api(self):
         with program_guard(Program(), Program()):
             repeat_times = [2, 2]
-            x1 = fluid.layers.data(name='x1', shape=[4], dtype="int32")
+            x1 = paddle.static.data(name='x1', shape=[-1, 4], dtype="int32")
             out = paddle.tile(x1, repeat_times)
             positive_2 = fluid.layers.fill_constant([1], dtype="int32", value=2)
             out2 = paddle.tile(x1, repeat_times=[positive_2, 2])
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_transpose_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_transpose_op_mlu.py
index a802f9da215b45e35636cf1a50f67d7c8ec508ea..60cd1e27c70ba36fd95800ffe68ba390cfacebcf 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_transpose_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_transpose_op_mlu.py
@@ -188,7 +188,7 @@ class TestTransposeOpError(unittest.TestCase):
     def test_errors(self):
         paddle.enable_static()
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[10, 5, 3], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 10, 5, 3], dtype='float32')
 
             def test_x_Variable_check():
                 # the Input(x)'s type must be Variable
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_where_index_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_where_index_op_mlu.py
index 877207eab7d2e620e1008814dfa44e3306680530..a12c4427a786d51d461befd98ad8903454398c60 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_where_index_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_where_index_op_mlu.py
@@ -107,7 +107,7 @@ class TestRank3(TestWhereIndexOp):
 class TestWhereOpError(unittest.TestCase):
     def test_api(self):
         with program_guard(Program(), Program()):
-            cond = fluid.layers.data(name='cond', shape=[4], dtype='bool')
+            cond = paddle.static.data(name='cond', shape=[-1, 4], dtype='bool')
             result = paddle.nonzero(cond)
 
             exe = fluid.Executor(paddle.device.MLUPlace(0))
diff --git a/python/paddle/fluid/tests/unittests/mlu/test_where_op_mlu.py b/python/paddle/fluid/tests/unittests/mlu/test_where_op_mlu.py
index a2d0c1a699dba8b5e98e5d9020266be0e66289b3..a2fc6e2073c667ceab1692aea26a5b3b7b84e3da 100644
--- a/python/paddle/fluid/tests/unittests/mlu/test_where_op_mlu.py
+++ b/python/paddle/fluid/tests/unittests/mlu/test_where_op_mlu.py
@@ -90,14 +90,14 @@ class TestWhereAPI(unittest.TestCase):
         for x_stop_gradient in [False, True]:
             for y_stop_gradient in [False, True]:
                 with fluid.program_guard(Program(), Program()):
-                    cond = fluid.layers.data(
-                        name='cond', shape=self.shape, dtype='bool'
+                    cond = paddle.static.data(
+                        name='cond', shape=[-1] + self.shape, dtype='bool'
                     )
-                    x = fluid.layers.data(
-                        name='x', shape=self.shape, dtype='float32'
+                    x = paddle.static.data(
+                        name='x', shape=[-1] + self.shape, dtype='float32'
                     )
-                    y = fluid.layers.data(
-                        name='y', shape=self.shape, dtype='float32'
+                    y = paddle.static.data(
+                        name='y', shape=[-1] + self.shape, dtype='float32'
                     )
                     x.stop_gradient = x_stop_gradient
                     y.stop_gradient = y_stop_gradient
@@ -137,8 +137,8 @@ class TestWhereAPI(unittest.TestCase):
     def test_api_broadcast(self, use_mlu=False):
         main_program = Program()
         with fluid.program_guard(main_program):
-            x = fluid.layers.data(name='x', shape=[4, 1], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[4, 2], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 4, 1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 4, 2], dtype='float32')
             x_i = np.array([[0.9383, 0.1983, 3.2, 1.2]]).astype('float32')
             y_i = np.array([[1.0, 1.0, 1.0, 1.0], [1.0, 1.0, 1.0, 1.0]]).astype(
                 'float32'
@@ -161,8 +161,8 @@ class TestWhereAPI(unittest.TestCase):
         main_program = Program()
         with fluid.program_guard(main_program):
             cond_shape = [2, 4]
-            cond = fluid.layers.data(
-                name='cond', shape=cond_shape, dtype='bool'
+            cond = paddle.static.data(
+                name='cond', shape=[-1] + cond_shape, dtype='bool'
             )
             x_data = 1.0
             y_data = 2.0
@@ -185,11 +185,11 @@ class TestWhereAPI(unittest.TestCase):
         paddle.enable_static()
         main_program = Program()
         with fluid.program_guard(main_program):
-            cond = fluid.layers.data(
-                name='cond', shape=cond_shape, dtype='bool'
+            cond = paddle.static.data(
+                name='cond', shape=[-1] + cond_shape, dtype='bool'
             )
-            x = fluid.layers.data(name='x', shape=x_shape, dtype='float32')
-            y = fluid.layers.data(name='y', shape=y_shape, dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1] + x_shape, dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1] + y_shape, dtype='float32')
             cond_data_tmp = np.random.random(size=cond_shape).astype('float32')
             cond_data = cond_data_tmp < 0.3
             x_data = np.random.random(size=x_shape).astype('float32')
@@ -340,7 +340,7 @@ class TestWhereDygraphAPI(unittest.TestCase):
     def test_where_condition(self):
         data = np.array([[True, False], [False, True]])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[(-1), 2])
+            x = paddle.static.data(name='x', shape=[(-1), 2])
             y = paddle.where(x)
             self.assertEqual(type(y), tuple)
             self.assertEqual(len(y), 2)
@@ -353,7 +353,7 @@ class TestWhereDygraphAPI(unittest.TestCase):
         np.testing.assert_allclose(expect_out, np.array(res))
         data = np.array([True, True, False])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[(-1)])
+            x = paddle.static.data(name='x', shape=[(-1)])
             y = paddle.where(x)
             self.assertEqual(type(y), tuple)
             self.assertEqual(len(y), 1)
@@ -379,9 +379,9 @@ class TestWhereOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_Variable)
 
             def test_type():
-                x = fluid.layers.data(name='x', shape=[4], dtype='bool')
-                y = fluid.layers.data(name='y', shape=[4], dtype='float16')
-                cond = fluid.layers.data(name='cond', shape=[4], dtype='int32')
+                x = paddle.static.data(name='x', shape=[-1, 4], dtype='bool')
+                y = paddle.static.data(name='y', shape=[-1, 4], dtype='float16')
+                cond = paddle.static.data(name='cond', shape=[-1, 4], dtype='int32')
                 paddle.where(cond, x, y)
 
             self.assertRaises(TypeError, test_type)
diff --git a/python/paddle/fluid/tests/unittests/npu/collective_identity_op_npu.py b/python/paddle/fluid/tests/unittests/npu/collective_identity_op_npu.py
index f8f8612ca89e32ceee64e1f1ab7acd69c80612d3..c0a871e51f6737c603e976aeba30cd32b192cda7 100644
--- a/python/paddle/fluid/tests/unittests/npu/collective_identity_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/collective_identity_op_npu.py
@@ -42,8 +42,8 @@ class TestCollectiveIdentity(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             toutdata = main_prog.current_block().create_var(
                 name="outofgather",
diff --git a/python/paddle/fluid/tests/unittests/npu/sync_batch_norm_op_npu.py b/python/paddle/fluid/tests/unittests/npu/sync_batch_norm_op_npu.py
index 72c822ad3a58e5e9f80ac811749c335b998855b0..0d0b469881894541bae31fc9805d4de099f520e7 100644
--- a/python/paddle/fluid/tests/unittests/npu/sync_batch_norm_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/sync_batch_norm_op_npu.py
@@ -74,11 +74,10 @@ class TestSyncBatchNormOpTraining(TestSyncBatchNormRunnerBase):
         use_cudnn = False
         with fluid.unique_name.guard():
             with fluid.program_guard(main, startup):
-                data = fluid.layers.data(
+                data = paddle.static.data(
                     name='input',
-                    shape=self.dshape,
+                    shape=[-1] + self.dshape,
                     dtype=self.dtype,
-                    append_batch_size=False,
                 )
                 conv = paddle.static.nn.conv2d(
                     input=data,
diff --git a/python/paddle/fluid/tests/unittests/npu/test_atan_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_atan_op_npu.py
index fe22453f4fa45f71912681676fbea290245b6804..c6fa750a17b7a8a7bb6b30b075576ff60eb519d2 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_atan_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_atan_op_npu.py
@@ -52,7 +52,7 @@ class TestAtan(OpTest):
     def test_out_name(self):
         with fluid.program_guard(fluid.Program()):
             np_x = np.array([0.1])
-            data = fluid.layers.data(name="X", shape=[1])
+            data = paddle.static.data(name="X", shape=[1])
             out = paddle.atan(data, name='Y')
             place = paddle.NPUPlace(0)
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_clip_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_clip_op_npu.py
index 122429a7f8454cd687b90b3e503b531727d478f0..37dcf8465bc2193367810d3aec791a22657fb138 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_clip_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_clip_op_npu.py
@@ -122,7 +122,7 @@ class TestClipOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_Variable)
 
             def test_dtype():
-                x2 = fluid.layers.data(name='x2', shape=[1], dtype='int32')
+                x2 = paddle.static.data(name='x2', shape=[-1, 1], dtype='int32')
                 paddle.clip(x=x2, min=-1.0, max=1.0)
 
             self.assertRaises(TypeError, test_dtype)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py
index 4fd0be14c12b54b22178543eb5dddf8df481dfb5..6caa442fc05357b542c13d7f7612b79a4e7b3484 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_compare_op_npu.py
@@ -44,9 +44,9 @@ def create_test_class(op_type, typename, callback):
         def test_errors(self):
             paddle.enable_static()
             with program_guard(Program(), Program()):
-                a = fluid.layers.data(name='a', shape=[2], dtype='float32')
-                b = fluid.layers.data(name='b', shape=[2], dtype='float32')
-                c = fluid.layers.data(name='c', shape=[2], dtype='int16')
+                a = paddle.static.data(name='a', shape=[-1, 2], dtype='float32')
+                b = paddle.static.data(name='b', shape=[-1, 2], dtype='float32')
+                c = paddle.static.data(name='c', shape=[-1, 2], dtype='int16')
                 d = fluid.create_lod_tensor(np.array([[-1]]), [[1]], self.place)
 
                 op = eval("fluid.layers.%s" % self.op_type)
@@ -134,8 +134,8 @@ def create_test_class(op_type, typename, callback):
         def test_attr_name(self):
             paddle.enable_static()
             with program_guard(Program(), Program()):
-                x = fluid.layers.data(name='x', shape=[4], dtype=typename)
-                y = fluid.layers.data(name='y', shape=[4], dtype=typename)
+                x = paddle.static.data(name='x', shape=[-1, 4], dtype=typename)
+                y = paddle.static.data(name='y', shape=[-1, 4], dtype=typename)
                 op = eval("paddle.%s" % (self.op_type))
                 out = op(x=x, y=y, name="name_%s" % (self.op_type))
             self.assertEqual("name_%s" % (self.op_type) in out.name, True)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_conv2d_transpose_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_conv2d_transpose_op_npu.py
index aae34ebfb5be22c44f5ce3c56c4fcb65b7269a61..b2f966629dad38001ebb23298b73d886ac6a2318 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_conv2d_transpose_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_conv2d_transpose_op_npu.py
@@ -429,11 +429,11 @@ class TestWithEvenUpsample_NHWC_output_padding_FP16(
 
 class TestConv2DTransposeAPI(unittest.TestCase):
     def test_case1(self):
-        data1 = fluid.layers.data(
-            name='data1', shape=[3, 5, 5], dtype='float32'
+        data1 = paddle.static.data(
+            name='data1', shape=[-1, 3, 5, 5], dtype='float32'
         )
-        data2 = fluid.layers.data(
-            name='data2', shape=[5, 5, 3], dtype='float32'
+        data2 = paddle.static.data(
+            name='data2', shape=[-1, 5, 5, 3], dtype='float32'
         )
         out1 = paddle.static.nn.conv2d_transpose(
             input=data1,
diff --git a/python/paddle/fluid/tests/unittests/npu/test_conv3d_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_conv3d_op_npu.py
index 65ec9a489f4a532b4d79755aed10e688b9eee51f..cdb7def1c242737bcdaacf2d5c526c1e5942899f 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_conv3d_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_conv3d_op_npu.py
@@ -350,17 +350,15 @@ class TestCase1_AsyPadding(TestConv3DOp_2):
 class TestConv3DAPI(unittest.TestCase):
     def test_api(self):
 
-        input_NDHWC = fluid.layers.data(
+        input_NDHWC = paddle.static.data(
             name="input_NDHWC",
             shape=[2, 5, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NCDHW = fluid.layers.data(
+        input_NCDHW = paddle.static.data(
             name="input_NCDHW",
             shape=[2, 3, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -433,10 +431,9 @@ class TestConv3DAPI(unittest.TestCase):
 
 class TestConv3DAPI_Error(unittest.TestCase):
     def test_api(self):
-        input = fluid.layers.data(
+        input = paddle.static.data(
             name="input",
             shape=[2, 5, 5, 5, 4],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -519,10 +516,9 @@ class TestConv3DAPI_Error(unittest.TestCase):
         self.assertRaises(ValueError, run_5)
 
         # ValueError: channel dimmention
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[2, 5, 5, 5, -1],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/npu/test_expand_as_v2_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_expand_as_v2_op_npu.py
index ca2b0195b4280e2c739f51339e3d87115e464673..9625f72d8cff816062e3bd28de006e7e80d7f8a9 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_expand_as_v2_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_expand_as_v2_op_npu.py
@@ -122,14 +122,13 @@ class TestExpandAsV2API(unittest.TestCase):
     def test_api(self):
         input1 = np.random.random([12, 14]).astype("float32")
         input2 = np.random.random([2, 12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
+        x = paddle.static.data(
+            name='x', shape=[12, 14], dtype="float32"
         )
 
-        y = fluid.layers.data(
+        y = paddle.static.data(
             name='target_tensor',
             shape=[2, 12, 14],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/npu/test_expand_v2_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_expand_v2_op_npu.py
index 0ea52f04d9d7693465de77209f20049da59a98a6..2176f67047e8b4ccd88924265de69b9eb09712e3 100755
--- a/python/paddle/fluid/tests/unittests/npu/test_expand_v2_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_expand_v2_op_npu.py
@@ -252,9 +252,9 @@ class TestExpandV2Error(unittest.TestCase):
             )
             shape = [2, 2]
             self.assertRaises(TypeError, paddle.tensor.expand, x1, shape)
-            x2 = fluid.layers.data(name='x2', shape=[2], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 2], dtype="uint8")
             self.assertRaises(TypeError, paddle.tensor.expand, x2, shape)
-            x3 = fluid.layers.data(name='x3', shape=[2], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 2], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tensor.expand, x3, shape)
 
@@ -264,18 +264,16 @@ class TestExpandV2API(unittest.TestCase):
     def test_static(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
             input = np.random.random([12, 14]).astype("float32")
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name='x',
                 shape=[12, 14],
-                append_batch_size=False,
                 dtype="float32",
             )
 
             positive_2 = fluid.layers.fill_constant([1], "int32", 12)
-            expand_shape = fluid.layers.data(
+            expand_shape = paddle.static.data(
                 name="expand_shape",
                 shape=[2],
-                append_batch_size=False,
                 dtype="int32",
             )
 
diff --git a/python/paddle/fluid/tests/unittests/npu/test_gather_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_gather_op_npu.py
index a6fa001076c2a53c7f6ed842c9fd04ffc5036d52..3ce9042d75fbda9a7a8b521d076efb0250ccb78f 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_gather_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_gather_op_npu.py
@@ -84,8 +84,8 @@ class TestCase1(TestGatherOp):
 class API_TestGather(unittest.TestCase):
     def test_out1(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[-1, 2], dtype='float32')
-            index = fluid.layers.data('index', shape=[-1, 1], dtype='int32')
+            data1 = paddle.static.data('data1', shape=[-1, 2], dtype='float32')
+            index = paddle.static.data('index', shape=[-1, 1], dtype='int32')
             out = paddle.gather(data1, index)
             place = paddle.NPUPlace(0)
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_group_norm_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_group_norm_op_npu.py
index 7f95e2b55c66df89ffefc9dfa151eb06428f3bd7..56430ee7c13aacec87d0acc2731b5f16c53fcd0a 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_group_norm_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_group_norm_op_npu.py
@@ -57,8 +57,8 @@ class TestGroupNormOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_x_type)
 
             def test_x_dtype():
-                x2 = fluid.layers.data(
-                    name='x2', shape=[2, 100, 3, 5], dtype='int32'
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 2, 100, 3, 5], dtype='int32'
                 )
                 groups = 2
                 paddle.static.nn.group_norm(x2, groups)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_momentum_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_momentum_op_npu.py
index b500b44e7e57cc0812b9572f005fa61e8c8e27e9..440f681b018a74c102951117822fa7a64432221f 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_momentum_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_momentum_op_npu.py
@@ -108,8 +108,8 @@ class TestMomentumV2(unittest.TestCase):
         place = fluid.NPUPlace(0)
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
             cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
@@ -236,8 +236,8 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
         place = fluid.NPUPlace(0)
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
             cost =paddle.nn.functional.square_error_cost(input=y_predict, label=y)
             avg_cost = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_one_hot_v2_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_one_hot_v2_op_npu.py
index 4ccd33134bed7e5b2036baf6ce7f37dbb8ce2726..d50c6dea782419c6b57fed9829b9ae7a7b95715f 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_one_hot_v2_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_one_hot_v2_op_npu.py
@@ -224,7 +224,7 @@ class TestOneHotOpApi(unittest.TestCase):
             )
 
     def _run(self, depth):
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         one_hot_label = fluid.one_hot(input=label, depth=depth)
 
         place = fluid.NPUPlace(0)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_sin_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_sin_op_npu.py
index 971184956e6d8bb81d48cd5544af02ec54971cb0..b441efd013f59276b0d4452164158c7b12c3cbb5 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_sin_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_sin_op_npu.py
@@ -55,7 +55,7 @@ def test_class(op_type, typename):
         def test_out_name(self):
             with fluid.program_guard(fluid.Program()):
                 np_x = np.array([0.1])
-                data = fluid.layers.data(name="X", shape=[1])
+                data = paddle.static.data(name="X", shape=[-1, 1])
                 out = eval("paddle.%s(data, name='Y')" % self.op_type)
                 place = fluid.NPUPlace(0)
                 exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_split_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_split_op_npu.py
index d6a6cf9d573232aed6f3e5a8e657bb666fd4c875..38cc028760a3d2a1bf03e10b12b1fce0e25e4f3c 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_split_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_split_op_npu.py
@@ -100,7 +100,7 @@ class TestCase5(TestCase1):
 class API_TestSplit(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float32')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float32')
             x0, x1 = paddle.split(data, num_or_sections=(3, 7), axis=1)
             place = fluid.NPUPlace(0)
             exe = fluid.Executor(place)
@@ -114,7 +114,7 @@ class API_TestSplit(unittest.TestCase):
 class API_TestSplit2(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float32')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float32')
             x0, x1 = paddle.split(data, num_or_sections=2, axis=1)
             place = fluid.NPUPlace(0)
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_stack_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_stack_op_npu.py
index d14b84b645e74c5887f9ff7e6f88647f1bd139c7..1e185195d140adf83478d77d4a2d27ecf06d4d0a 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_stack_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_stack_op_npu.py
@@ -196,9 +196,9 @@ class TestTensorStackAPIWithLoDTensorArray(unittest.TestCase):
 class API_test(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[1, 2], dtype='float32')
-            data2 = fluid.layers.data('data2', shape=[1, 2], dtype='float32')
-            data3 = fluid.layers.data('data3', shape=[1, 2], dtype='float32')
+            data1 = paddle.static.data('data1', shape=[-1, 1, 2], dtype='float32')
+            data2 = paddle.static.data('data2', shape=[-1, 1, 2], dtype='float32')
+            data3 = paddle.static.data('data3', shape=[-1, 1, 2], dtype='float32')
             result_stack = paddle.stack([data1, data2, data3], axis=0)
             place = paddle.NPUPlace(0)
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/npu/test_strided_slice_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_strided_slice_op_npu.py
index c4470b101d18435919fcc6b9a3038c7e5df544bb..1bee4627e16ac11680466372ea516cabdb265199 100755
--- a/python/paddle/fluid/tests/unittests/npu/test_strided_slice_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_strided_slice_op_npu.py
@@ -580,20 +580,19 @@ class TestStridedSliceAPI(unittest.TestCase):
         input = np.random.random([3, 4, 5, 6]).astype("float64")
         minus_1 = fluid.layers.fill_constant([1], "int32", -1)
         minus_3 = fluid.layers.fill_constant([1], "int32", -3)
-        starts = fluid.layers.data(
-            name='starts', shape=[3], dtype='int32', append_batch_size=False
+        starts = paddle.static.data(
+            name='starts', shape=[3], dtype='int32'
         )
-        ends = fluid.layers.data(
-            name='ends', shape=[3], dtype='int32', append_batch_size=False
+        ends = paddle.static.data(
+            name='ends', shape=[3], dtype='int32'
         )
-        strides = fluid.layers.data(
-            name='strides', shape=[3], dtype='int32', append_batch_size=False
+        strides = paddle.static.data(
+            name='strides', shape=[3], dtype='int32'
         )
 
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[3, 4, 5, 6],
-            append_batch_size=False,
             dtype="float64",
         )
         out_1 = paddle.strided_slice(
diff --git a/python/paddle/fluid/tests/unittests/npu/test_sync_batch_norm_op_npu_extra.py b/python/paddle/fluid/tests/unittests/npu/test_sync_batch_norm_op_npu_extra.py
index 5c8d00b8fc272aca00b92be60ed9cee001dfc909..9cd59bb8a5c75b2d7f47b0800f0e34a527451ffe 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_sync_batch_norm_op_npu_extra.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_sync_batch_norm_op_npu_extra.py
@@ -46,7 +46,7 @@ class TestDygraphSyncBatchNormAPIError(unittest.TestCase):
 
             # the input dtype of SyncBatchNorm must be float16 or float32
             # float16 only can be set on GPU place and NPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32")
             self.assertRaises(TypeError, my_sync_batch_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/npu/test_where_index_npu.py b/python/paddle/fluid/tests/unittests/npu/test_where_index_npu.py
index 315f475a5c5bfd28699041c5bd6de919e0e07a22..a2bb351d6549b3253dacad5259a336a1b18d8f1a 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_where_index_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_where_index_npu.py
@@ -97,7 +97,7 @@ class TestRank3(TestWhereIndexOp):
 class TestWhereOpError(unittest.TestCase):
     def test_api(self):
         with program_guard(Program(), Program()):
-            cond = fluid.layers.data(name='cond', shape=[4], dtype='bool')
+            cond = paddle.static.data(name='cond', shape=[-1, 4], dtype='bool')
             result = paddle.nonzero(cond)
 
             exe = fluid.Executor(paddle.NPUPlace(0))
diff --git a/python/paddle/fluid/tests/unittests/npu/test_where_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_where_op_npu.py
index 9e1126d0aa0436204f28e904f3e8f4cd439dfc77..5e9baa696fc368fffb85abe8f8b8643cd0c84008 100755
--- a/python/paddle/fluid/tests/unittests/npu/test_where_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_where_op_npu.py
@@ -133,8 +133,8 @@ class TestNPUWhereAPI(unittest.TestCase):
         train_prog = fluid.Program()
         startup = fluid.Program()
         with fluid.program_guard(train_prog, startup):
-            x = fluid.layers.data(name='x', shape=[4, 1], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[4, 2], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 4, 1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 4, 2], dtype='float32')
             x_i = np.array([[0.9383, 0.1983, 3.2, 1.2]]).astype("float32")
             y_i = np.array([[1.0, 1.0, 1.0, 1.0], [1.0, 1.0, 1.0, 1.0]]).astype(
                 "float32"
diff --git a/python/paddle/fluid/tests/unittests/npu/test_while_op_npu.py b/python/paddle/fluid/tests/unittests/npu/test_while_op_npu.py
index 8314f836c7cc03afae360cc0d0bfb751c7118343..a8bcfeae68d3f618b2480734c7e16e0bb3b5f414 100644
--- a/python/paddle/fluid/tests/unittests/npu/test_while_op_npu.py
+++ b/python/paddle/fluid/tests/unittests/npu/test_while_op_npu.py
@@ -27,14 +27,14 @@ paddle.enable_static()
 
 class TestWhileOp(unittest.TestCase):
     def simple_net(self):
-        d0 = layers.data(
-            "d0", shape=[10], append_batch_size=False, dtype='float32'
+        d0 = paddle.static.data(
+            "d0", shape=[10], dtype='float32'
         )
-        d1 = layers.data(
-            "d1", shape=[10], append_batch_size=False, dtype='float32'
+        d1 = paddle.static.data(
+            "d1", shape=[10], dtype='float32'
         )
-        d2 = layers.data(
-            "d2", shape=[10], append_batch_size=False, dtype='float32'
+        d2 = paddle.static.data(
+            "d2", shape=[10], dtype='float32'
         )
         # fill_constant npu op doesn't support int64
         i = layers.zeros(shape=[1], dtype='int32')
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_concat.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_concat.py
index cb262f9495f8fd81cb471e4f9f7082f0b5e8d1b1..bd18a205259a3958484f1f27426c8e8d6d7dfc0f 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_concat.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_concat.py
@@ -89,7 +89,9 @@ class TestSequenceConcatOpError(unittest.TestCase):
     def test_errors(self):
         def test_input_list():
             # the input type must be list
-            x_data = fluid.layers.data(name='x', shape=[4], dtype='float32')
+            x_data = paddle.static.data(
+                name='x', shape=[-1, 4], dtype='float32'
+            )
             fluid.layers.sequence_concat(input=x_data)
 
         self.assertRaises(TypeError, test_input_list)
@@ -97,12 +99,16 @@ class TestSequenceConcatOpError(unittest.TestCase):
         def test_variable1():
             # the input element type must be Variable
             x1_data = np.array([[3, 5]]).astype('float32')
-            y1_data = fluid.layers.data(name='y1', shape=[4], dtype='float32')
+            y1_data = paddle.static.data(
+                name='y1', shape=[-1, 4], dtype='float32'
+            )
             fluid.layers.sequence_concat(input=[x1_data, y1_data])
 
         def test_variable2():
             x2_data = np.array([[3, 5]]).astype('float32')
-            y2_data = fluid.layers.data(name='y2', shape=[4], dtype='float32')
+            y2_data = paddle.static.data(
+                name='y2', shape=[-1, 4], dtype='float32'
+            )
             fluid.layers.sequence_concat(input=[y2_data, x2_data])
 
         for i in range(2):
@@ -113,8 +119,12 @@ class TestSequenceConcatOpError(unittest.TestCase):
 
         def test_dtype():
             # dtype must be 'float32', 'float64', 'int64'
-            x3_data = fluid.layers.data(name="x3", shape=[3, 5], dtype='int32')
-            y3_data = fluid.layers.data(name="y3", shape=[3, 5], dtype='int16')
+            x3_data = paddle.static.data(
+                name="x3", shape=[-1, 3, 5], dtype='int32'
+            )
+            y3_data = paddle.static.data(
+                name="y3", shape=[-1, 3, 5], dtype='int16'
+            )
             input_list = [x3_data, y3_data]
             fluid.layers.sequence_concat(input=input_list)
 
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_conv.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_conv.py
index ec296d969c8492cf5a72a6f1ea61a3016d3c8caa..2376716b7ca5bdc4f1d435f3ac9d775ac01ba4fe 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_conv.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_conv.py
@@ -18,6 +18,8 @@ import unittest
 
 import numpy as np
 
+import paddle
+
 sys.path.append("../")
 from op_test import OpTest
 
@@ -283,7 +285,7 @@ class TestSeqConvApi(unittest.TestCase):
     def test_api(self):
         import paddle.fluid as fluid
 
-        x = fluid.layers.data('x', shape=[32], lod_level=1)
+        x = paddle.static.data('x', shape=[-1, 32], lod_level=1)
         y = fluid.layers.sequence_conv(
             input=x, num_filters=2, filter_size=3, padding_start=None
         )
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_first_step.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_first_step.py
index 85e225c896cb52d6b25ff73176375bc5839d637f..704462b040a51fdf247406afc327a31798b23c31 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_first_step.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_first_step.py
@@ -16,6 +16,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 from paddle.fluid.framework import Program, program_guard
 
@@ -35,10 +36,9 @@ class TestSequenceFirstStepOpError(unittest.TestCase):
 
             def test_input_dtype():
                 # the dtype of input must be int64
-                type_data = fluid.layers.data(
+                type_data = paddle.static.data(
                     name='type_data',
                     shape=[7, 1],
-                    append_batch_size=False,
                     dtype='int64',
                     lod_level=1,
                 )
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_last_step.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_last_step.py
index e269b46517ecab33bab858c8501f455751c02997..165f941f58576fe540e2aed6cc4608f5178a4aeb 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_last_step.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_last_step.py
@@ -16,6 +16,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 from paddle.fluid.framework import Program, program_guard
 
@@ -35,10 +36,9 @@ class TestSequenceLastStepOpError(unittest.TestCase):
 
             def test_input_dtype():
                 # the dtype of input must be int64
-                type_data = fluid.layers.data(
+                type_data = paddle.static.data(
                     name='type_data',
                     shape=[7, 1],
-                    append_batch_size=False,
                     dtype='int64',
                     lod_level=1,
                 )
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_pad_op.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_pad_op.py
index a3dd8373d98fc586fafb576090a7f3b78767de9a..421b4e96893558c7f18b37838ac7bf4f125aae26 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_pad_op.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_pad_op.py
@@ -17,6 +17,8 @@ import unittest
 
 import numpy as np
 
+import paddle
+
 sys.path.append("../")
 from op_test import OpTest
 
@@ -162,8 +164,8 @@ class TestSequencePadOpError(unittest.TestCase):
         self.assertRaises(TypeError, test_x_variable)
 
         def test_pad_value_variable():
-            x1 = fluid.layers.data(
-                name='x1', shape=[10, 5], dtype='float32', lod_level=1
+            x1 = paddle.static.data(
+                name='x1', shape=[-1, 10, 5], dtype='float32', lod_level=1
             )
             pad_value1 = np.array([0.0], dtype=np.float32)
             fluid.layers.sequence_pad(x=x1, pad_value=pad_value1)
@@ -171,8 +173,8 @@ class TestSequencePadOpError(unittest.TestCase):
         self.assertRaises(TypeError, test_pad_value_variable)
 
         def test_dtype():
-            x2 = fluid.layers.data(
-                name='x2', shape=[10, 5], dtype='int16', lod_level=1
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 10, 5], dtype='int16', lod_level=1
             )
             pad_value2 = fluid.layers.assign(
                 input=np.array([0.0], dtype=np.int32)
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_reshape.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_reshape.py
index 5cda07af7de90094948c8276c61c8f9481347e69..9f903251507eba0c3981a1fabb02a45a119bb9b0 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_reshape.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_reshape.py
@@ -17,6 +17,8 @@ import unittest
 
 import numpy as np
 
+import paddle
+
 sys.path.append("../")
 from op_test import OpTest
 
@@ -93,10 +95,9 @@ class TestSequenceReshapeOpError(unittest.TestCase):
         self.assertRaises(TypeError, test_variable)
 
         def test_dtype():
-            x1 = fluid.layers.data(
+            x1 = paddle.static.data(
                 name='x1',
-                shape=[2, 6],
-                append_batch_size=False,
+                shape=[-1, 2, 6],
                 dtype='float16',
                 lod_level=1,
             )
diff --git a/python/paddle/fluid/tests/unittests/sequence/test_sequence_reverse.py b/python/paddle/fluid/tests/unittests/sequence/test_sequence_reverse.py
index 13897df00ab939c8478caa3accc97525b910e4be..f141806aac2339ee2fd7a78531e03d8bcbbb07af 100644
--- a/python/paddle/fluid/tests/unittests/sequence/test_sequence_reverse.py
+++ b/python/paddle/fluid/tests/unittests/sequence/test_sequence_reverse.py
@@ -17,6 +17,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 
 sys.path.append("../")
@@ -106,7 +107,9 @@ class TestSequenceReverseOpError(unittest.TestCase):
 
         def test_dtype():
             # dtype must be 'float32', 'float64', 'int8', 'int32', 'int64'
-            x2_data = fluid.layers.data(name='x2', shape=[4], dtype='float16')
+            x2_data = paddle.static.data(
+                name='x2', shape=[-1, 4], dtype='float16'
+            )
             fluid.layers.sequence_reverse(x=x2_data)
 
         self.assertRaises(TypeError, test_dtype)
diff --git a/python/paddle/fluid/tests/unittests/seresnext_net.py b/python/paddle/fluid/tests/unittests/seresnext_net.py
index beb24de94677e755a91ee1b0dbbb4989833be73f..6e1ad62dc13f5f04ffc3b6dd340b56d90bdec9b0 100644
--- a/python/paddle/fluid/tests/unittests/seresnext_net.py
+++ b/python/paddle/fluid/tests/unittests/seresnext_net.py
@@ -131,8 +131,10 @@ img_shape = [3, 224, 224]
 
 def SE_ResNeXt50Small(use_feed):
 
-    img = fluid.layers.data(name='image', shape=img_shape, dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(
+        name='image', shape=[-1] + img_shape, dtype='float32'
+    )
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     conv = conv_bn_layer(
         input=img, num_filters=16, filter_size=3, stride=2, act='relu'
diff --git a/python/paddle/fluid/tests/unittests/simple_nets.py b/python/paddle/fluid/tests/unittests/simple_nets.py
index 2b41107061d07c3fb5fd1f0b5d29495f138973b6..a3ff2b686574477e8f12132c97c1c9e0e538cf5b 100644
--- a/python/paddle/fluid/tests/unittests/simple_nets.py
+++ b/python/paddle/fluid/tests/unittests/simple_nets.py
@@ -40,8 +40,8 @@ def simple_fc_net_with_inputs(img, label, class_num=10):
 
 
 def simple_fc_net(use_feed=None):
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     return simple_fc_net_with_inputs(img, label, class_num=10)
 
 
@@ -70,8 +70,8 @@ def batchnorm_fc_with_inputs(img, label, class_num=10):
 
 
 def fc_with_batchnorm(use_feed=None):
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     return batchnorm_fc_with_inputs(img, label, class_num=10)
 
 
@@ -89,10 +89,10 @@ def bow_net(
     This model is from https://github.com/PaddlePaddle/models:
     fluid/PaddleNLP/text_classification/nets.py
     """
-    data = fluid.layers.data(
-        name="words", shape=[1], dtype="int64", lod_level=1
+    data = paddle.static.data(
+        name="words", shape=[-1, 1], dtype="int64", lod_level=1
     )
-    label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+    label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
     emb = fluid.layers.embedding(
         input=data, is_sparse=is_sparse, size=[dict_dim, emb_dim]
     )
diff --git a/python/paddle/fluid/tests/unittests/test_accuracy_op.py b/python/paddle/fluid/tests/unittests/test_accuracy_op.py
index 446240e9fb71da1d63187311f205e01442c62dfd..431d8b24bcee2f7581c25a542a71fc9cb1a6340f 100755
--- a/python/paddle/fluid/tests/unittests/test_accuracy_op.py
+++ b/python/paddle/fluid/tests/unittests/test_accuracy_op.py
@@ -66,16 +66,18 @@ class TestAccuracyOpError(unittest.TestCase):
             x1 = fluid.create_lod_tensor(
                 np.array([[-1]]), [[1]], fluid.CPUPlace()
             )
-            label = fluid.layers.data(
+            label = paddle.static.data(
                 name='label', shape=[-1, 1], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.static.accuracy, x1, label)
             self.assertRaises(TypeError, paddle.metric.accuracy, x1, label)
             # The input dtype of accuracy_op must be float32 or float64.
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="int32")
             self.assertRaises(TypeError, paddle.static.accuracy, x2, label)
             self.assertRaises(TypeError, paddle.metric.accuracy, x2, label)
-            x3 = fluid.layers.data(name='input', shape=[-1, 2], dtype="float16")
+            x3 = paddle.static.data(
+                name='input', shape=[-1, 2], dtype="float16"
+            )
             paddle.static.accuracy(input=x3, label=label)
             paddle.metric.accuracy(input=x3, label=label)
 
diff --git a/python/paddle/fluid/tests/unittests/test_activation_nn_grad.py b/python/paddle/fluid/tests/unittests/test_activation_nn_grad.py
index 48349cfe910b35f35e1e19399c39bd40a0b63726..8333da1accfda292da640353f0f005bd29224873 100644
--- a/python/paddle/fluid/tests/unittests/test_activation_nn_grad.py
+++ b/python/paddle/fluid/tests/unittests/test_activation_nn_grad.py
@@ -21,7 +21,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 import paddle.nn.functional as F
 
 
@@ -31,7 +30,7 @@ class TestSigmoidTripleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = F.sigmoid(x)
         x_arr = np.random.random(shape).astype(dtype)
@@ -58,7 +57,7 @@ class TestSigmoidDoubleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = F.sigmoid(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -88,7 +87,7 @@ class TestTanhTripleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.tanh(x)
         x_arr = np.random.random(shape).astype(dtype)
@@ -118,7 +117,7 @@ class TestTanhDoubleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.tanh(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -148,7 +147,7 @@ class TestAbsDoubleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.abs(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -176,7 +175,7 @@ class TestReluDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
         y = F.relu(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -206,7 +205,7 @@ class TestLeakyReluDoubleGradCheck(unittest.TestCase):
         alpha = 0.2
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
 
         y = paddle.nn.functional.leaky_relu(x, alpha)
@@ -241,7 +240,7 @@ class TestELUDoubleGradCheck(unittest.TestCase):
         dtype = np.float64
         SEED = 0
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
 
         y = paddle.nn.functional.elu(x, alpha=alpha)
@@ -275,7 +274,7 @@ class TestCELUDoubleGradCheck(unittest.TestCase):
         dtype = np.float64
         SEED = 0
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
 
         y = F.celu(x, alpha=alpha)
@@ -307,7 +306,7 @@ class TestSqrtDoubleGradCheck(unittest.TestCase):
         eps = 0.0001
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
 
         y = paddle.sqrt(x)
@@ -339,7 +338,7 @@ class TestRsqrtDoubleGradCheck(unittest.TestCase):
         eps = 0.0001
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
 
         y = paddle.rsqrt(x)
@@ -372,7 +371,7 @@ class TestSquareDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
         y = paddle.square(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -403,7 +402,7 @@ class TestLogDoubleGradCheck(unittest.TestCase):
         eps = 1e-6
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
         y = paddle.log(x)
 
@@ -434,7 +433,7 @@ class TestSinDoubleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.sin(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -464,7 +463,7 @@ class TestCosDoubleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.cos(x)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -494,7 +493,7 @@ class TestPowDoubleGradCheck1(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 1e-6
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.pow(x, 2)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -523,7 +522,7 @@ class TestPowDoubleGradCheck2(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 1e-6
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.pow(x, 1)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -552,7 +551,7 @@ class TestSinTripleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.sin(x)
         x_arr = np.random.random(shape).astype(dtype)
@@ -582,7 +581,7 @@ class TestPowTripleGradCheck1(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 1e-6
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.pow(x, 1)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -611,7 +610,7 @@ class TestPowTripleGradCheck2(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 1e-6
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.pow(x, 2)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -640,7 +639,7 @@ class TestPowTripleGradCheck3(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 1e-6
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.pow(x, 4)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -669,7 +668,7 @@ class TestCosTripleGradCheck(unittest.TestCase):
         shape = [2, 3, 7, 9]
         eps = 0.0005
         dtype = np.float64
-        x = layers.data('x', shape, False, dtype=dtype)
+        x = paddle.static.data('x', shape, dtype=dtype)
         x.persistable = True
         y = paddle.cos(x)
         x_arr = np.random.random(shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_activation_op.py b/python/paddle/fluid/tests/unittests/test_activation_op.py
index 2f0c594746a508f1c2794c381c5d35658d315887..1f0a49cbb35ae875f14cbd93406dc6bd3fa887d2 100755
--- a/python/paddle/fluid/tests/unittests/test_activation_op.py
+++ b/python/paddle/fluid/tests/unittests/test_activation_op.py
@@ -34,13 +34,13 @@ class TestSqrtOpError(unittest.TestCase):
             in1 = 1
             self.assertRaises(TypeError, paddle.sqrt, in1)
             # The input dtype of sqrt op must be float16, float32, float64.
-            in2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            in2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.sqrt, in2)
 
-            in3 = fluid.layers.data(
-                name='input3', shape=[12, 10], dtype="float16"
+            in3 = paddle.static.data(
+                name='input3', shape=[-1, 12, 10], dtype="float16"
             )
             paddle.sqrt(x=in3)
 
@@ -167,8 +167,8 @@ class TestExpm1API(unittest.TestCase):
 class TestParameter:
     def test_out_name(self):
         with fluid.program_guard(fluid.Program()):
-            np_x = np.array([0.1])
-            data = fluid.layers.data(name="X", shape=[1])
+            np_x = np.array([0.1]).astype('float32').reshape((-1, 1))
+            data = paddle.static.data(name="X", shape=[-1, 1], dtype="float32")
             out = eval("paddle.%s(data, name='Y')" % self.op_type)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
@@ -520,8 +520,8 @@ class TestAtan(TestActivation, TestParameter):
 
     def test_out_name(self):
         with fluid.program_guard(fluid.Program()):
-            np_x = np.array([0.1])
-            data = fluid.layers.data(name="X", shape=[1])
+            np_x = np.array([0.1]).astype('float32').reshape((-1, 1))
+            data = paddle.static.data(name="X", shape=[-1, 1], dtype="float32")
             out = paddle.atan(data, name='Y')
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
@@ -582,10 +582,9 @@ class TestSinhAPI(unittest.TestCase):
             input_x = np.random.uniform(0.1, 1, test_data_shape).astype(
                 "float32"
             )
-            data_x = fluid.layers.data(
+            data_x = paddle.static.data(
                 name="data_x",
                 shape=test_data_shape,
-                append_batch_size=False,
                 dtype="float32",
             )
 
@@ -667,10 +666,9 @@ class TestCoshAPI(unittest.TestCase):
             input_x = np.random.uniform(0.1, 1, test_data_shape).astype(
                 "float32"
             )
-            data_x = fluid.layers.data(
+            data_x = paddle.static.data(
                 name="data_x",
                 shape=test_data_shape,
-                append_batch_size=False,
                 dtype="float32",
             )
 
@@ -2399,12 +2397,8 @@ class TestLog(TestActivation):
         self.check_grad(['X'], 'Out', check_eager=True)
 
     def test_error(self):
-        in1 = fluid.layers.data(
-            name="in1", shape=[11, 17], append_batch_size=False, dtype="int32"
-        )
-        in2 = fluid.layers.data(
-            name="in2", shape=[11, 17], append_batch_size=False, dtype="int64"
-        )
+        in1 = paddle.static.data(name="in1", shape=[11, 17], dtype="int32")
+        in2 = paddle.static.data(name="in2", shape=[11, 17], dtype="int64")
 
         self.assertRaises(TypeError, paddle.log, in1)
         self.assertRaises(TypeError, paddle.log, in2)
@@ -2569,10 +2563,9 @@ class TestLog1pAPI(unittest.TestCase):
     def test_api(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
             input_x = np.random.uniform(0.1, 1, [11, 17]).astype("float64")
-            data_x = fluid.layers.data(
+            data_x = paddle.static.data(
                 name="data_x",
                 shape=[11, 17],
-                append_batch_size=False,
                 dtype="float64",
             )
 
@@ -2718,12 +2711,8 @@ class TestPow_factor_tensor(TestActivation):
 
     def test_api(self):
         input = np.random.uniform(1, 2, [11, 17]).astype("float32")
-        x = fluid.layers.data(
-            name="x", shape=[11, 17], append_batch_size=False, dtype="float32"
-        )
-        res = fluid.layers.data(
-            name="res", shape=[11, 17], append_batch_size=False, dtype="float32"
-        )
+        x = paddle.static.data(name="x", shape=[11, 17], dtype="float32")
+        res = paddle.static.data(name="res", shape=[11, 17], dtype="float32")
 
         factor_1 = 2.0
         factor_2 = fluid.layers.fill_constant([1], "float32", 3.0)
diff --git a/python/paddle/fluid/tests/unittests/test_adadelta_op.py b/python/paddle/fluid/tests/unittests/test_adadelta_op.py
index 95a485ce3a4e7d4acd0146eacfa33e10116de262..51435ccb95fc998dd4013b4a08d06ff3fc8ff3c6 100644
--- a/python/paddle/fluid/tests/unittests/test_adadelta_op.py
+++ b/python/paddle/fluid/tests/unittests/test_adadelta_op.py
@@ -143,8 +143,8 @@ class TestAdadeltaV2(unittest.TestCase):
         place = fluid.CPUPlace()
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1)
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
diff --git a/python/paddle/fluid/tests/unittests/test_addmm_op.py b/python/paddle/fluid/tests/unittests/test_addmm_op.py
index 9a1385c63b5dbc39e51025a13feae6532cd3c758..2e4a9515b6aef1442b9391d9aad531c04e822206 100644
--- a/python/paddle/fluid/tests/unittests/test_addmm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_addmm_op.py
@@ -76,109 +76,86 @@ class TestAddMMOpError(unittest.TestCase):
             self.assertRaises(TypeError, paddle.addmm, input, x1, x2)
 
             # The input dtype of mul_op must be float32 or float64.
-            input = fluid.layers.data(
+            input = paddle.static.data(
                 name='input',
                 shape=[4, 4],
                 dtype="int32",
-                append_batch_size=False,
-            )
-            x3 = fluid.layers.data(
-                name='x3', shape=[4, 4], dtype="int32", append_batch_size=False
-            )
-            x4 = fluid.layers.data(
-                name='x4', shape=[4, 4], dtype="int32", append_batch_size=False
             )
+            x3 = paddle.static.data(name='x3', shape=[4, 4], dtype="int32")
+            x4 = paddle.static.data(name='x4', shape=[4, 4], dtype="int32")
             self.assertRaises(TypeError, paddle.addmm, input, x3, x4)
             # x and y dimension mismatch
-            x5 = fluid.layers.data(
+            x5 = paddle.static.data(
                 name='x5',
                 shape=[4, 5],
                 dtype="float32",
-                append_batch_size=False,
             )
-            x6 = fluid.layers.data(
+            x6 = paddle.static.data(
                 name='x6',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
             self.assertRaises(ValueError, paddle.addmm, input, x5, x6)
             # input and x are not broadcastable
-            x7 = fluid.layers.data(
+            x7 = paddle.static.data(
                 name='x7',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
-            x8 = fluid.layers.data(
+            x8 = paddle.static.data(
                 name='x8',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
-            input1 = fluid.layers.data(
+            input1 = paddle.static.data(
                 name='input1',
                 shape=[2, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
             self.assertRaises(ValueError, paddle.addmm, input1, x7, x8)
             # input and x are not broadcastable
-            x9 = fluid.layers.data(
+            x9 = paddle.static.data(
                 name='x9',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
-            x10 = fluid.layers.data(
+            x10 = paddle.static.data(
                 name='x10',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
-            input2 = fluid.layers.data(
+            input2 = paddle.static.data(
                 name='input2',
                 shape=[1, 2],
                 dtype="float32",
-                append_batch_size=False,
             )
             self.assertRaises(ValueError, paddle.addmm, input2, x9, x10)
-            x11 = fluid.layers.data(
+            x11 = paddle.static.data(
                 name='x11',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
-            )
-            x12 = fluid.layers.data(
-                name='x12',
-                shape=[4, 4],
-                dtype="float32",
-                append_batch_size=False,
             )
-            input3 = fluid.layers.data(
+            x12 = paddle.static.data(name='x12', shape=[4, 4], dtype="float32")
+            input3 = paddle.static.data(
                 name='input3',
                 shape=[4, 2],
                 dtype="float32",
-                append_batch_size=False,
             )
             self.assertRaises(ValueError, paddle.addmm, input3, x11, x12)
-            x13 = fluid.layers.data(
+            x13 = paddle.static.data(
                 name='x13',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
-            x14 = fluid.layers.data(
+            x14 = paddle.static.data(
                 name='x14',
                 shape=[4, 4],
                 dtype="float32",
-                append_batch_size=False,
             )
-            input4 = fluid.layers.data(
+            input4 = paddle.static.data(
                 name='input4',
                 shape=[3, 1],
                 dtype="float32",
-                append_batch_size=False,
             )
             self.assertRaises(ValueError, paddle.addmm, input4, x13, x14)
 
diff --git a/python/paddle/fluid/tests/unittests/test_argsort_op.py b/python/paddle/fluid/tests/unittests/test_argsort_op.py
index 17614692f69d8ac035dee1272ef928a0bd3d39ef..e3f90d7fd2d1001ecaf219201bcf462f7e0b06f7 100644
--- a/python/paddle/fluid/tests/unittests/test_argsort_op.py
+++ b/python/paddle/fluid/tests/unittests/test_argsort_op.py
@@ -84,13 +84,17 @@ class TestArgsortOpCPU(unittest.TestCase):
         )
 
         with fluid.program_guard(self.main_program, self.startup_program):
-            x = fluid.layers.data(
-                name="x", shape=self.input_shape, dtype=self.dtype
+            x = paddle.static.data(
+                name="x", shape=[-1] + list(self.input_shape), dtype=self.dtype
             )
             x.stop_gradient = False
-            label = fluid.layers.data(
-                name="label", shape=self.input_shape, dtype=self.dtype
+            x.desc.set_need_check_feed(False)
+            label = paddle.static.data(
+                name="label",
+                shape=[-1] + list(self.input_shape),
+                dtype=self.dtype,
             )
+            label.desc.set_need_check_feed(False)
             self.index = paddle.argsort(
                 x=x, axis=self.axis, descending=self.descending
             )
diff --git a/python/paddle/fluid/tests/unittests/test_array_read_write_op.py b/python/paddle/fluid/tests/unittests/test_array_read_write_op.py
index 3e4543cd1ce8ddf1d3b83a80663c0d784f02bb1b..497dc31477aeca69ba64f5b465e234b0d7b03694 100644
--- a/python/paddle/fluid/tests/unittests/test_array_read_write_op.py
+++ b/python/paddle/fluid/tests/unittests/test_array_read_write_op.py
@@ -62,9 +62,9 @@ class TestArrayReadWrite(unittest.TestCase):
     def test_read_write(self):
         paddle.enable_static()
         x = [
-            layers.data(name='x0', shape=[100]),
-            layers.data(name='x1', shape=[100]),
-            layers.data(name='x2', shape=[100]),
+            paddle.static.data(name='x0', shape=[-1, 100]),
+            paddle.static.data(name='x1', shape=[-1, 100]),
+            paddle.static.data(name='x2', shape=[-1, 100]),
         ]
         for each_x in x:
             each_x.stop_gradient = False
diff --git a/python/paddle/fluid/tests/unittests/test_assign_op.py b/python/paddle/fluid/tests/unittests/test_assign_op.py
index 02f649b39bfab231fe0691c39b217e017ebf3a9a..8017840c50c115304776c9d1b4973ab97bca493f 100644
--- a/python/paddle/fluid/tests/unittests/test_assign_op.py
+++ b/python/paddle/fluid/tests/unittests/test_assign_op.py
@@ -22,7 +22,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 from paddle.fluid.backward import append_backward
 
@@ -261,7 +260,7 @@ class TestAssignDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [3, 4, 5], False, dtype)
+        data = paddle.static.data('data', [3, 4, 5], dtype)
         data.persistable = True
         out = paddle.fluid.layers.assign(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -292,7 +291,7 @@ class TestAssignTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [3, 4, 5], False, dtype)
+        data = paddle.static.data('data', [3, 4, 5], dtype)
         data.persistable = True
         out = paddle.fluid.layers.assign(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_async_ssa_graph_executor_mnist.py b/python/paddle/fluid/tests/unittests/test_async_ssa_graph_executor_mnist.py
index fb6dc442a73c92f12051e20613aa44eeb77e22a3..3c6594e659e894e3d1f96c7a142e84463d9877a8 100644
--- a/python/paddle/fluid/tests/unittests/test_async_ssa_graph_executor_mnist.py
+++ b/python/paddle/fluid/tests/unittests/test_async_ssa_graph_executor_mnist.py
@@ -26,8 +26,10 @@ BATCH_SIZE = 64
 
 def convolutional_neural_network(use_py_reader):
     with fluid.unique_name.guard():
-        img = fluid.layers.data(name='img', shape=[1, 28, 28], dtype='float32')
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        img = paddle.static.data(
+            name='img', shape=[-1, 1, 28, 28], dtype='float32'
+        )
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         py_reader = None
         if use_py_reader:
diff --git a/python/paddle/fluid/tests/unittests/test_auto_parallel_mapper.py b/python/paddle/fluid/tests/unittests/test_auto_parallel_mapper.py
index 946b8959d0b4d397a451cb8a14bb7a8fa0ca11fb..25e4ab9aa8b4a48c9f4ab6349e5ba4bbb00cd67a 100644
--- a/python/paddle/fluid/tests/unittests/test_auto_parallel_mapper.py
+++ b/python/paddle/fluid/tests/unittests/test_auto_parallel_mapper.py
@@ -38,7 +38,7 @@ from paddle.distributed.auto_parallel.parallelizer import AutoParallelizer
 from paddle.distributed.auto_parallel.partitioner import Partitioner
 from paddle.distributed.auto_parallel.reshard import Resharder
 from paddle.distributed.fleet import auto
-from paddle.fluid import core, layers
+from paddle.fluid import core
 from paddle.fluid.initializer import NumpyArrayInitializer
 
 if os.getenv("CUDA_VISIBLE_DEVICES") is not None:
@@ -588,7 +588,9 @@ class TestAutoParallelMapper(unittest.TestCase):
         root_id = 0
         nranks = 2
         with fluid.program_guard(train_program, startup_program):
-            input = layers.data(name="input", shape=[10, 10], dtype='float32')
+            input = paddle.static.data(
+                name="input", shape=[-1, 10, 10], dtype='float32'
+            )
             output = train_program.current_block().create_var(
                 name="outofbroadcast",
                 dtype='float32',
diff --git a/python/paddle/fluid/tests/unittests/test_batch_norm_op.py b/python/paddle/fluid/tests/unittests/test_batch_norm_op.py
index 6802a8a9ea995726eed7985b7bea78e2e1eb910a..c2a6c468e5c8f184e6e74a98151d6c008ba997d4 100644
--- a/python/paddle/fluid/tests/unittests/test_batch_norm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_batch_norm_op.py
@@ -763,7 +763,9 @@ class TestBatchNormOpError(unittest.TestCase):
 
             # the input dtype of batch_norm must be float16 or float32 or float64
             # float16 only can be set on GPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
+            )
             self.assertRaises(TypeError, paddle.static.nn.batch_norm, x2)
 
 
@@ -779,7 +781,9 @@ class TestDygraphBatchNormAPIError(unittest.TestCase):
 
             # the input dtype of BatchNorm must be float16 or float32 or float64
             # float16 only can be set on GPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
+            )
             self.assertRaises(TypeError, batch_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_bmm_op.py b/python/paddle/fluid/tests/unittests/test_bmm_op.py
index 5269f27ccdaa3990cd5a425286bf56947c0eb4cc..aaa0e1f97fde11959ce938dca1ceb88cc502a89c 100644
--- a/python/paddle/fluid/tests/unittests/test_bmm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_bmm_op.py
@@ -41,10 +41,10 @@ class TestBmmOp(OpTest):
 class API_TestBmm(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data(
+            data1 = paddle.static.data(
                 'data1', shape=[-1, 3, 4], dtype='float64'
             )
-            data2 = fluid.layers.data(
+            data2 = paddle.static.data(
                 'data2', shape=[-1, 4, 5], dtype='float64'
             )
             result_bmm = paddle.bmm(data1, data2)
diff --git a/python/paddle/fluid/tests/unittests/test_boxps.py b/python/paddle/fluid/tests/unittests/test_boxps.py
index 4f2d3ee138a77573fad6a64ae53c99f0bc18935a..b6a1a845aa326e3b455a76596a3898e5fc5feef3 100644
--- a/python/paddle/fluid/tests/unittests/test_boxps.py
+++ b/python/paddle/fluid/tests/unittests/test_boxps.py
@@ -96,11 +96,11 @@ class TestPullBoxSparseOP(unittest.TestCase):
         paddle.enable_static()
         program = fluid.Program()
         with fluid.program_guard(program):
-            x = fluid.layers.data(
-                name='x', shape=[1], dtype='int64', lod_level=0
+            x = paddle.static.data(
+                name='x', shape=[-1, 1], dtype='int64', lod_level=0
             )
-            y = fluid.layers.data(
-                name='y', shape=[1], dtype='int64', lod_level=0
+            y = paddle.static.data(
+                name='y', shape=[-1, 1], dtype='int64', lod_level=0
             )
             emb_x, emb_y = _pull_box_sparse([x, y], size=1)
 
diff --git a/python/paddle/fluid/tests/unittests/test_broadcast_tensors_op.py b/python/paddle/fluid/tests/unittests/test_broadcast_tensors_op.py
index 8a1b6a52a2baa92412b4830df600aaef4a795f4a..6eec711c49e0abb854cd5da84216e7a966221ec6 100644
--- a/python/paddle/fluid/tests/unittests/test_broadcast_tensors_op.py
+++ b/python/paddle/fluid/tests/unittests/test_broadcast_tensors_op.py
@@ -164,11 +164,11 @@ class TestBroadcastTensorsAPI(unittest.TestCase):
     def test_api(self):
         def test_static():
             inputs = [
-                paddle.fluid.layers.data(
-                    shape=[4, 1, 4, 1], dtype='float32', name="x0"
+                paddle.static.data(
+                    shape=[-1, 4, 1, 4, 1], dtype='float32', name="x0"
                 ),
-                paddle.fluid.layers.data(
-                    shape=[1, 4, 1, 4], dtype='float32', name="x1"
+                paddle.static.data(
+                    shape=[-1, 1, 4, 1, 4], dtype='float32', name="x1"
                 ),
             ]
             paddle.broadcast_tensors(inputs)
@@ -196,33 +196,33 @@ class TestRaiseBroadcastTensorsError(unittest.TestCase):
     def test_errors(self):
         def test_type():
             inputs = [
-                paddle.fluid.layers.data(
-                    shape=[1, 1, 1, 1], dtype='float32', name="x4"
+                paddle.static.data(
+                    shape=[-1, 1, 1, 1, 1], dtype='float32', name="x4"
                 ),
-                paddle.fluid.layers.data(
-                    shape=[1, 4, 1, 1], dtype='float64', name="x5"
+                paddle.static.data(
+                    shape=[-1, 1, 4, 1, 1], dtype='float64', name="x5"
                 ),
             ]
             paddle.broadcast_tensors(inputs)
 
         def test_dtype():
             inputs = [
-                paddle.fluid.layers.data(
-                    shape=[1, 1, 1, 1], dtype='int8', name="x6"
+                paddle.static.data(
+                    shape=[-1, 1, 1, 1, 1], dtype='int8', name="x6"
                 ),
-                paddle.fluid.layers.data(
-                    shape=[1, 4, 1, 1], dtype='int8', name="x7"
+                paddle.static.data(
+                    shape=[-1, 1, 4, 1, 1], dtype='int8', name="x7"
                 ),
             ]
             paddle.broadcast_tensors(inputs)
 
         def test_bcast_semantics():
             inputs = [
-                paddle.fluid.layers.data(
-                    shape=[1, 3, 1, 1], dtype='float32', name="x9"
+                paddle.static.data(
+                    shape=[-1, 1, 3, 1, 1], dtype='float32', name="x9"
                 ),
-                paddle.fluid.layers.data(
-                    shape=[1, 8, 1, 1], dtype='float32', name="x10"
+                paddle.static.data(
+                    shape=[-1, 1, 8, 1, 1], dtype='float32', name="x10"
                 ),
             ]
             paddle.broadcast_tensors(inputs)
diff --git a/python/paddle/fluid/tests/unittests/test_broadcast_to_op.py b/python/paddle/fluid/tests/unittests/test_broadcast_to_op.py
index ea8e7648b1d32d3a35ee91fc4e7a4fb167da36a7..897c5b54b169b7e6054cf6692b951f4d8d234ecc 100644
--- a/python/paddle/fluid/tests/unittests/test_broadcast_to_op.py
+++ b/python/paddle/fluid/tests/unittests/test_broadcast_to_op.py
@@ -31,9 +31,9 @@ class TestBroadcastToError(unittest.TestCase):
             )
             shape = [2, 2]
             self.assertRaises(TypeError, paddle.tensor.broadcast_to, x1, shape)
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="uint8")
             self.assertRaises(TypeError, paddle.tensor.broadcast_to, x2, shape)
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tensor.broadcast_to, x3, shape)
 
@@ -42,15 +42,12 @@ class TestBroadcastToError(unittest.TestCase):
 class TestBroadcastToAPI(unittest.TestCase):
     def test_api(self):
         input = np.random.random([12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
-        )
+        x = paddle.static.data(name='x', shape=[12, 14], dtype="float32")
 
         positive_2 = fluid.layers.fill_constant([1], "int32", 12)
-        expand_shape = fluid.layers.data(
+        expand_shape = paddle.static.data(
             name="expand_shape",
             shape=[2],
-            append_batch_size=False,
             dtype="int32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/test_cast_op.py b/python/paddle/fluid/tests/unittests/test_cast_op.py
index ea7f24a4883867704c8a793d154d1b354a4e3dee..0e451c3e9fe776c45629b61499c4fc5e5880c91d 100644
--- a/python/paddle/fluid/tests/unittests/test_cast_op.py
+++ b/python/paddle/fluid/tests/unittests/test_cast_op.py
@@ -22,7 +22,6 @@ from op_test import OpTest, convert_float_to_uint16, convert_uint16_to_float
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 
 
@@ -142,7 +141,7 @@ class TestCastDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3, 4], False, dtype)
+        data = paddle.static.data('data', [2, 3, 4], dtype)
         data.persistable = True
         out = paddle.cast(data, 'float64')
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -173,7 +172,7 @@ class TestCastTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3, 4], False, dtype)
+        data = paddle.static.data('data', [2, 3, 4], dtype)
         data.persistable = True
         out = paddle.cast(data, 'float64')
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_communicator_async.py b/python/paddle/fluid/tests/unittests/test_communicator_async.py
index ebc65cd5ac240b5ea80abdefa6b188b7d3a0840e..322ec2f9205a01cc4bb129cb35a551de02986cfc 100644
--- a/python/paddle/fluid/tests/unittests/test_communicator_async.py
+++ b/python/paddle/fluid/tests/unittests/test_communicator_async.py
@@ -27,8 +27,8 @@ import paddle.fluid as fluid
 
 class TestCommunicator(unittest.TestCase):
     def net(self):
-        x = fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/test_communicator_geo.py b/python/paddle/fluid/tests/unittests/test_communicator_geo.py
index 9019e9e9e3f5ecb110c9cd5200ae8e6906306990..0e836dca1c2e515d95b9ff6c94f4383203779a5f 100644
--- a/python/paddle/fluid/tests/unittests/test_communicator_geo.py
+++ b/python/paddle/fluid/tests/unittests/test_communicator_geo.py
@@ -31,8 +31,10 @@ paddle.enable_static()
 
 class TestCommunicatorGeoEnd2End(unittest.TestCase):
     def net(self):
-        x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-        x1 = fluid.layers.data(name='x1', shape=[1], dtype='int64', lod_level=1)
+        x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+        x1 = paddle.static.data(
+            name='x1', shape=[-1, 1], dtype='int64', lod_level=1
+        )
 
         emb = fluid.layers.embedding(
             input=x1,
@@ -47,8 +49,7 @@ class TestCommunicatorGeoEnd2End(unittest.TestCase):
         pool = fluid.layers.sequence_pool(input=emb, pool_type="sum")
         z = fluid.layers.concat(input=[x, pool], axis=1)
         y_predict = paddle.static.nn.fc(x=z, size=1)
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
-
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         return avg_cost, x, x1, y
diff --git a/python/paddle/fluid/tests/unittests/test_communicator_ps_gpu.py b/python/paddle/fluid/tests/unittests/test_communicator_ps_gpu.py
index f0c1e9c8850d8547094562ad6e1ace089e0a121e..b767d98281ed182ecc9679bd3db3a063885f1335 100644
--- a/python/paddle/fluid/tests/unittests/test_communicator_ps_gpu.py
+++ b/python/paddle/fluid/tests/unittests/test_communicator_ps_gpu.py
@@ -51,8 +51,8 @@ class TestCommunicator(unittest.TestCase):
         role = role_maker.PaddleCloudRoleMaker()
 
         fleet.init(role)
-        x = fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         slots_vars = [x, y]
 
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
diff --git a/python/paddle/fluid/tests/unittests/test_compare_op.py b/python/paddle/fluid/tests/unittests/test_compare_op.py
index fa98771ce1ab5eb0522720adc2dcfd7d9915bdd1..55cc9d933fc02bdd079f51b13a1f1a8492fbd8f9 100755
--- a/python/paddle/fluid/tests/unittests/test_compare_op.py
+++ b/python/paddle/fluid/tests/unittests/test_compare_op.py
@@ -41,9 +41,9 @@ def create_test_class(op_type, typename, callback):
         def test_errors(self):
             paddle.enable_static()
             with program_guard(Program(), Program()):
-                x = fluid.layers.data(name='x', shape=[2], dtype='int32')
-                y = fluid.layers.data(name='y', shape=[2], dtype='int32')
-                a = fluid.layers.data(name='a', shape=[2], dtype='int16')
+                x = paddle.static.data(name='x', shape=[-1, 2], dtype='int32')
+                y = paddle.static.data(name='y', shape=[-1, 2], dtype='int32')
+                a = paddle.static.data(name='a', shape=[-1, 2], dtype='int16')
                 op = eval("paddle.%s" % self.op_type)
                 self.assertRaises(TypeError, op, x=x, y=a)
                 self.assertRaises(TypeError, op, x=a, y=y)
@@ -415,8 +415,8 @@ def create_paddle_case(op_type, callback):
         def test_attr_name(self):
             paddle.enable_static()
             with program_guard(Program(), Program()):
-                x = fluid.layers.data(name='x', shape=[4], dtype='int32')
-                y = fluid.layers.data(name='y', shape=[4], dtype='int32')
+                x = paddle.static.data(name='x', shape=[-1, 4], dtype='int32')
+                y = paddle.static.data(name='y', shape=[-1, 4], dtype='int32')
                 op = eval("paddle.%s" % (self.op_type))
                 out = op(x=x, y=y, name="name_%s" % (self.op_type))
             self.assertEqual("name_%s" % (self.op_type) in out.name, True)
@@ -439,7 +439,7 @@ class TestCompareOpError(unittest.TestCase):
         paddle.enable_static()
         with program_guard(Program(), Program()):
             # The input x and y of compare_op must be Variable.
-            x = fluid.layers.data(name='x', shape=[1], dtype="float32")
+            x = paddle.static.data(name='x', shape=[-1, 1], dtype="float32")
             y = fluid.create_lod_tensor(
                 numpy.array([[-1]]), [[1]], fluid.CPUPlace()
             )
diff --git a/python/paddle/fluid/tests/unittests/test_compiled_program.py b/python/paddle/fluid/tests/unittests/test_compiled_program.py
index 28c3feb010e817924d945760c975bb241e88c69d..a2ea57f83576ab338cf42d274101e742fa91b60c 100644
--- a/python/paddle/fluid/tests/unittests/test_compiled_program.py
+++ b/python/paddle/fluid/tests/unittests/test_compiled_program.py
@@ -102,10 +102,10 @@ class TestCompiledProgramError(unittest.TestCase):
         self.assertRaises(TypeError, fluid.CompiledProgram, "program")
 
     def build_simple_model(self):
-        img = fluid.layers.data(
-            name='image', shape=[1, 28, 28], dtype='float32'
+        img = paddle.static.data(
+            name='image', shape=[-1, 1, 28, 28], dtype='float32'
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         prediction = paddle.static.nn.fc(x=img, size=10, activation='softmax')
         loss = paddle.nn.functional.cross_entropy(
             input=prediction, label=label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_concat_op.py b/python/paddle/fluid/tests/unittests/test_concat_op.py
index d20b3e92c45f18f33ead2356ff13390bc6501c4c..7d7c2ddf70d49dcd385d9a760f8d4b10e9535a63 100644
--- a/python/paddle/fluid/tests/unittests/test_concat_op.py
+++ b/python/paddle/fluid/tests/unittests/test_concat_op.py
@@ -20,7 +20,6 @@ from decorator_helper import prog_scope
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, core, program_guard
 from paddle.fluid.tests.unittests.op_test import (
     OpTest,
@@ -250,7 +249,7 @@ class TestConcatOpError(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
             # The input type of concat_op should be list.
-            x1 = fluid.layers.data(shape=[4], dtype='int32', name='x1')
+            x1 = paddle.static.data(shape=[-1, 4], dtype='int32', name='x1')
             fluid.layers.concat(x1)
             # The item in input must be Variable.
             x2 = fluid.create_lod_tensor(
@@ -261,12 +260,12 @@ class TestConcatOpError(unittest.TestCase):
             )
             self.assertRaises(TypeError, fluid.layers.concat, [x2])
             # The input dtype of concat_op must be float16, float32, float64, int32, int64.
-            x4 = fluid.layers.data(shape=[4], dtype='uint8', name='x4')
-            x5 = fluid.layers.data(shape=[4], dtype='uint8', name='x5')
+            x4 = paddle.static.data(shape=[-1, 4], dtype='uint8', name='x4')
+            x5 = paddle.static.data(shape=[-1, 4], dtype='uint8', name='x5')
             self.assertRaises(TypeError, fluid.layers.concat, [x4, x5])
-            x6 = fluid.layers.data(shape=[4], dtype='float16', name='x6')
-            x7 = fluid.layers.data(shape=[4], dtype='float16', name='x7')
-            x8 = fluid.layers.data(shape=[4], dtype='float32', name='x8')
+            x6 = paddle.static.data(shape=[-1, 4], dtype='float16', name='x6')
+            x7 = paddle.static.data(shape=[-1, 4], dtype='float16', name='x7')
+            x8 = paddle.static.data(shape=[-1, 4], dtype='float32', name='x8')
             fluid.layers.concat([x6, x7])
 
             # The type of axis in concat_op should be int or Variable.
@@ -369,9 +368,9 @@ class TestConcatAPI(unittest.TestCase):
             self.assertRaises(TypeError, fluid.layers.concat, [x4, x5])
 
             # The type of axis in concat_op should be int or Variable.
-            x6 = fluid.layers.data(shape=[4], dtype='float16', name='x6')
-            x7 = fluid.layers.data(shape=[4], dtype='float16', name='x7')
-            x8 = fluid.layers.data(shape=[4], dtype='float32', name='x8')
+            x6 = paddle.static.data(shape=[-1, 4], dtype='float16', name='x6')
+            x7 = paddle.static.data(shape=[-1, 4], dtype='float16', name='x7')
+            x8 = paddle.static.data(shape=[-1, 4], dtype='float32', name='x8')
 
             def test_axis_type():
                 paddle.concat([x6, x7], 3.2)
@@ -457,9 +456,9 @@ class TestConcatDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data1 = layers.data('data1', [2, 3], False, dtype)
+        data1 = paddle.static.data('data1', [2, 3], dtype)
         data1.persistable = True
-        data2 = layers.data('data2', [2, 3], False, dtype)
+        data2 = paddle.static.data('data2', [2, 3], dtype)
         data2.persistable = True
         out = paddle.concat([data1, data2])
         data1_arr = np.random.uniform(-1, 1, data1.shape).astype(dtype)
@@ -498,9 +497,9 @@ class TestConcatTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data1 = layers.data('data1', [2, 3, 4], False, dtype)
+        data1 = paddle.static.data('data1', [2, 3, 4], dtype)
         data1.persistable = True
-        data2 = layers.data('data2', [2, 3, 4], False, dtype)
+        data2 = paddle.static.data('data2', [2, 3, 4], dtype)
         data2.persistable = True
         out = paddle.concat([data1, data2], 1)
         data1_arr = np.random.uniform(-1, 1, data1.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_conditional_block.py b/python/paddle/fluid/tests/unittests/test_conditional_block.py
index 0aee7cadd7e52354b0c43fd3654a029b24ce6eb2..0f8f6b32c9a8c34ee183a86a696a32846e3d8eaa 100644
--- a/python/paddle/fluid/tests/unittests/test_conditional_block.py
+++ b/python/paddle/fluid/tests/unittests/test_conditional_block.py
@@ -30,7 +30,7 @@ class ConditionalBlockTest(unittest.TestCase):
         main_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(main_program, startup_program):
-            data = layers.data(name='X', shape=[1], dtype='float32')
+            data = paddle.static.data(name='X', shape=[-1, 1], dtype='float32')
             data.stop_gradient = False
             cond = ConditionalBlock(inputs=[data])
             out = paddle.tensor.create_tensor(dtype='float32')
diff --git a/python/paddle/fluid/tests/unittests/test_conv2d_api.py b/python/paddle/fluid/tests/unittests/test_conv2d_api.py
index 16bf938d1ca8abba4048a3b470f2288265bcea89..0c458c1fec484958842546617db4a4136a416234 100644
--- a/python/paddle/fluid/tests/unittests/test_conv2d_api.py
+++ b/python/paddle/fluid/tests/unittests/test_conv2d_api.py
@@ -26,17 +26,15 @@ import paddle.fluid.core as core
 class TestConv2DAPI(unittest.TestCase):
     def test_api(self):
 
-        input_NHWC = fluid.layers.data(
+        input_NHWC = paddle.static.data(
             name="input_NHWC",
             shape=[2, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NCHW = fluid.layers.data(
+        input_NCHW = paddle.static.data(
             name="input_NCHW",
             shape=[2, 3, 5, 5],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -120,10 +118,9 @@ class TestConv2DAPI(unittest.TestCase):
 
 class TestConv2DAPI_Error(unittest.TestCase):
     def test_api(self):
-        input = fluid.layers.data(
+        input = paddle.static.data(
             name="input",
             shape=[2, 5, 5, 5],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -206,10 +203,9 @@ class TestConv2DAPI_Error(unittest.TestCase):
         self.assertRaises(ValueError, run_5)
 
         # ValueError: channel dimmention
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[2, 5, 5, -1],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -293,10 +289,9 @@ class TestConv2DAPI_Error(unittest.TestCase):
         self.assertRaises(ValueError, run_10)
 
     def test_api_with_error_input(self):
-        input = fluid.layers.data(
+        input = paddle.static.data(
             name="error_input",
             shape=[1],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -325,9 +320,8 @@ class TestConv2DAPI_Error(unittest.TestCase):
 class TestConv2DEnviron(unittest.TestCase):
     def run1(self, place):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            inputs = fluid.layers.data(
+            inputs = paddle.static.data(
                 shape=[2, 3, 5, 5],
-                append_batch_size=False,
                 name="inputs",
                 dtype="float32",
             )
diff --git a/python/paddle/fluid/tests/unittests/test_conv2d_op.py b/python/paddle/fluid/tests/unittests/test_conv2d_op.py
index ca7017a4adeaf13f5a2db028a931b23dcf0bbba6..60c729b4f96d3ae78abedf5032cb09a35aab1537 100644
--- a/python/paddle/fluid/tests/unittests/test_conv2d_op.py
+++ b/python/paddle/fluid/tests/unittests/test_conv2d_op.py
@@ -713,8 +713,8 @@ class TestConv2DOpError(unittest.TestCase):
             def test_dtype():
                 # the input dtype of conv2d must be float16 or float32 or float64
                 # float16 only can be set on GPU place
-                x2 = fluid.layers.data(
-                    name='x2', shape=[3, 4, 5, 6], dtype="int32"
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                 )
                 paddle.static.nn.conv2d(x2, 1, 1)
 
diff --git a/python/paddle/fluid/tests/unittests/test_conv2d_transpose_op.py b/python/paddle/fluid/tests/unittests/test_conv2d_transpose_op.py
index 2d4694be2e9e343aefa784f1241942a2b316a092..afbce517f624373813e1896dcd2edc3594ca0f48 100644
--- a/python/paddle/fluid/tests/unittests/test_conv2d_transpose_op.py
+++ b/python/paddle/fluid/tests/unittests/test_conv2d_transpose_op.py
@@ -831,11 +831,11 @@ class TestCUDNNWithEvenUpsample_NHWC_FP16(TestCUDNN_FP16):
 
 class TestConv2DTransposeAPI(unittest.TestCase):
     def test_case1(self):
-        data1 = fluid.layers.data(
-            name='data1', shape=[3, 5, 5], dtype='float32'
+        data1 = paddle.static.data(
+            name='data1', shape=[-1, 3, 5, 5], dtype='float32'
         )
-        data2 = fluid.layers.data(
-            name='data2', shape=[5, 5, 3], dtype='float32'
+        data2 = paddle.static.data(
+            name='data2', shape=[-1, 5, 5, 3], dtype='float32'
         )
         out1 = paddle.static.nn.conv2d_transpose(
             input=data1,
@@ -918,7 +918,9 @@ class TestConv2DTransposeAPI(unittest.TestCase):
 
 class TestConv2DTransposeOpException(unittest.TestCase):
     def test_exception(self):
-        data = fluid.layers.data(name='data', shape=[3, 5, 5], dtype="float32")
+        data = paddle.static.data(
+            name='data', shape=[-1, 3, 5, 5], dtype="float32"
+        )
 
         def attr_data_format():
             out = paddle.static.nn.conv2d_transpose(
@@ -965,8 +967,8 @@ class TestConv2DTransposeOpException(unittest.TestCase):
 
         self.assertRaises(ValueError, attr_padding_with_data_format)
 
-        error_input = fluid.layers.data(
-            name='error_data', shape=[1], dtype="float32"
+        error_input = paddle.static.data(
+            name='error_data', shape=[-1, 1], dtype="float32"
         )
 
         def error_input_size():
diff --git a/python/paddle/fluid/tests/unittests/test_conv3d_op.py b/python/paddle/fluid/tests/unittests/test_conv3d_op.py
index dc2760307cfbd06f1f20c151649d5ed75ba05e33..4930fd25678c99a257c9aade0625cec35bfa1c2e 100644
--- a/python/paddle/fluid/tests/unittests/test_conv3d_op.py
+++ b/python/paddle/fluid/tests/unittests/test_conv3d_op.py
@@ -18,7 +18,6 @@ import numpy as np
 from op_test import OpTest
 
 import paddle
-import paddle.fluid as fluid
 import paddle.fluid.core as core
 
 
@@ -865,17 +864,15 @@ create_test_cudnn_channel_last_class(TestWith1x1_AsyPadding)
 class TestConv3DAPI(unittest.TestCase):
     def test_api(self):
 
-        input_NDHWC = fluid.layers.data(
+        input_NDHWC = paddle.static.data(
             name="input_NDHWC",
             shape=[2, 5, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NCDHW = fluid.layers.data(
+        input_NCDHW = paddle.static.data(
             name="input_NCDHW",
             shape=[2, 3, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -948,10 +945,9 @@ class TestConv3DAPI(unittest.TestCase):
 
 class TestConv3DAPI_Error(unittest.TestCase):
     def test_api(self):
-        input = fluid.layers.data(
+        input = paddle.static.data(
             name="input",
             shape=[2, 5, 5, 5, 4],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -1034,10 +1030,9 @@ class TestConv3DAPI_Error(unittest.TestCase):
         self.assertRaises(ValueError, run_5)
 
         # ValueError: channel dimmention
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[2, 5, 5, 5, -1],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/test_conv3d_transpose_part2_op.py b/python/paddle/fluid/tests/unittests/test_conv3d_transpose_part2_op.py
index 0ad217f6810243a0f134529d5b8ca0bf55aabcfc..d47f341652075478902c85294eb834c1dbf26663 100644
--- a/python/paddle/fluid/tests/unittests/test_conv3d_transpose_part2_op.py
+++ b/python/paddle/fluid/tests/unittests/test_conv3d_transpose_part2_op.py
@@ -86,11 +86,11 @@ class TestWithDilation_NHWC(TestConv3DTransposeOp):
 
 class TestConv3DTransposeAPI(unittest.TestCase):
     def test_case1(self):
-        data1 = fluid.layers.data(
-            name='data1', shape=[3, 5, 5, 5], dtype='float32'
+        data1 = paddle.static.data(
+            name='data1', shape=[-1, 3, 5, 5, 5], dtype='float32'
         )
-        data2 = fluid.layers.data(
-            name='data2', shape=[5, 5, 5, 3], dtype='float32'
+        data2 = paddle.static.data(
+            name='data2', shape=[-1, 5, 5, 5, 3], dtype='float32'
         )
 
         out1 = paddle.static.nn.conv3d_transpose(
@@ -174,8 +174,8 @@ class TestConv3DTransposeAPI(unittest.TestCase):
 
 class TestConv3DTransposeOpException(unittest.TestCase):
     def test_exception(self):
-        data = fluid.layers.data(
-            name='data', shape=[3, 5, 5, 5], dtype="float32"
+        data = paddle.static.data(
+            name='data', shape=[-1, 3, 5, 5, 5], dtype="float32"
         )
 
         def attr_data_format():
diff --git a/python/paddle/fluid/tests/unittests/test_conv_nn_grad.py b/python/paddle/fluid/tests/unittests/test_conv_nn_grad.py
index 2bb99a00be73a03a58a7c58448bfe96c006290a2..3dc153e58aaf2f9c7a020ac42fbc8dfbd1e332a2 100644
--- a/python/paddle/fluid/tests/unittests/test_conv_nn_grad.py
+++ b/python/paddle/fluid/tests/unittests/test_conv_nn_grad.py
@@ -21,7 +21,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 
 
 class TestConvDoubleGradCheck(unittest.TestCase):
@@ -30,7 +29,7 @@ class TestConvDoubleGradCheck(unittest.TestCase):
         shape = [2, 4, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(x, 2, 1, groups=1, bias_attr=False)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
 
@@ -57,7 +56,7 @@ class TestConvDoubleGradCheckTest0(unittest.TestCase):
         shape = [2, 4, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(x, 2, 1, bias_attr=False)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
 
@@ -83,7 +82,7 @@ class TestConvDoubleGradCheckTest1(unittest.TestCase):
         shape = [2, 3, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(x, 2, 1, padding=1, bias_attr=False)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
 
@@ -109,7 +108,7 @@ class TestConv3DDoubleGradCheck(unittest.TestCase):
         shape = [2, 4, 3, 4, 2]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(x, 2, 1, bias_attr=False)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
 
@@ -136,7 +135,7 @@ class TestConv3DDoubleGradCheckTest1(unittest.TestCase):
         shape = [2, 4, 5, 3, 2]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(x, 2, 1, padding=1, bias_attr=False)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
 
@@ -162,7 +161,7 @@ class TestConv2DoubleGradCheck_AsyPadding(unittest.TestCase):
         shape = [2, 2, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(
             input=x,
             num_filters=2,
@@ -195,7 +194,7 @@ class TestConv2DoubleGradCheck_PaddingSAME(unittest.TestCase):
         shape = [2, 2, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(
             input=x,
             num_filters=2,
@@ -228,7 +227,7 @@ class TestConv2DoubleGradCheck_PaddingVALID(unittest.TestCase):
         shape = [2, 2, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(
             input=x,
             num_filters=2,
@@ -261,7 +260,7 @@ class TestConv2DoubleGradCheck_ChannelLast(unittest.TestCase):
         shape = [2, 2, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(
             input=x,
             num_filters=2,
@@ -296,7 +295,7 @@ class TestConv2DoubleGradCheck_ChannelLast_AsyPadding(unittest.TestCase):
         shape = [2, 2, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d(
             input=x,
             num_filters=2,
@@ -331,7 +330,7 @@ class TestConv3DDoubleGradCheck_AsyPadding(unittest.TestCase):
         shape = [2, 2, 2, 2, 2]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(
             input=x,
             num_filters=2,
@@ -364,7 +363,7 @@ class TestConv3DoubleGradCheck_PaddingSAME(unittest.TestCase):
         shape = [2, 2, 2, 2, 2]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(
             input=x,
             num_filters=2,
@@ -398,7 +397,7 @@ class TestConv3DoubleGradCheck_PaddingVALID(unittest.TestCase):
         shape = [2, 2, 3, 3, 2]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(
             input=x,
             num_filters=2,
@@ -431,7 +430,7 @@ class TestConv3DDoubleGradCheck_ChannelLast(unittest.TestCase):
         shape = [2, 2, 2, 2, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(
             input=x,
             num_filters=2,
@@ -466,7 +465,7 @@ class TestConv3DDoubleGradCheck_ChannelLast_AsyPadding(unittest.TestCase):
         shape = [2, 2, 2, 2, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv3d(
             input=x,
             num_filters=2,
@@ -501,7 +500,7 @@ class TestDepthWiseConvDoubleGradCheck(unittest.TestCase):
         shape = [2, 4, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
 
         # condition of depthwise conv:
         # use_cudnn == False
@@ -538,8 +537,8 @@ class TestDepthWiseConvDoubleGradCheckCase1(unittest.TestCase):
         w_shape = [4, 1, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', x_shape, False, dtype)
-        w = layers.data('w', w_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
+        w = paddle.static.data('w', w_shape, dtype)
 
         # condition of depthwise conv:
         # use_cudnn == False
@@ -579,8 +578,8 @@ class TestConv3DDoubleGradCheck_NN(unittest.TestCase):
         w_shape = [6, 3, 3, 3, 3]
         eps = 0.005
         dtype = np.float32 if fluid.core.is_compiled_with_rocm() else np.float64
-        x = layers.data('x', x_shape, False, dtype)
-        w = layers.data('w', w_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
+        w = paddle.static.data('w', w_shape, dtype)
         x.persistable = True
         w.persistable = True
         y = paddle.nn.functional.conv3d(x, w)
diff --git a/python/paddle/fluid/tests/unittests/test_conv_transpose_nn_grad.py b/python/paddle/fluid/tests/unittests/test_conv_transpose_nn_grad.py
index 142359286bbea41b7fc2ce4bb5c0d8bd0ea4ffdb..34838dcd32c9d1722d1317352bb70495eaf5271d 100644
--- a/python/paddle/fluid/tests/unittests/test_conv_transpose_nn_grad.py
+++ b/python/paddle/fluid/tests/unittests/test_conv_transpose_nn_grad.py
@@ -21,7 +21,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 
 
 class TestConvTransposeDoubleGradCheck(unittest.TestCase):
@@ -35,7 +34,7 @@ class TestConvTransposeDoubleGradCheck(unittest.TestCase):
         dtype = np.float64
         if core.is_compiled_with_rocm():
             dtype = np.float32
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d_transpose(
             x, 2, filter_size=1, groups=1, bias_attr=False
         )
@@ -91,7 +90,7 @@ class TestConvTranspose2DoubleGradCheck_AsyPadding(
         dtype = np.float64
         if core.is_compiled_with_rocm():
             dtype = np.float32
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d_transpose(
             input=x,
             num_filters=2,
@@ -144,7 +143,7 @@ class TestConvTranspose2DoubleGradCheck_PaddingSAME(
         dtype = np.float64
         if core.is_compiled_with_rocm():
             dtype = np.float32
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d_transpose(
             input=x,
             num_filters=2,
@@ -197,7 +196,7 @@ class TestConvTranspose2DoubleGradCheck_PaddingVALID(
         dtype = np.float64
         if core.is_compiled_with_rocm():
             dtype = np.float32
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d_transpose(
             input=x,
             num_filters=2,
@@ -250,7 +249,7 @@ class TestConvTranspose2DoubleGradCheck_ChannelLast(
         dtype = np.float64
         if core.is_compiled_with_rocm():
             dtype = np.float32
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         y = paddle.static.nn.conv2d_transpose(
             input=x,
             num_filters=2,
diff --git a/python/paddle/fluid/tests/unittests/test_cross_entropy_op.py b/python/paddle/fluid/tests/unittests/test_cross_entropy_op.py
index a1f650dc63172de2dbad98d3362426a1b27f3366..ec5e5dbaa8a7f2f9043565f937706b70f3c1287f 100644
--- a/python/paddle/fluid/tests/unittests/test_cross_entropy_op.py
+++ b/python/paddle/fluid/tests/unittests/test_cross_entropy_op.py
@@ -429,11 +429,11 @@ class TestCrossEntropyOpError(unittest.TestCase):
             def test_dtype():
                 # the input dtype of cross_entropy must be float16 or float32 or float64
                 # float16 only can be set on GPU place
-                x2 = fluid.layers.data(
-                    name='x2', shape=[3, 4, 5, 6], dtype="int32"
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                 )
-                lab2 = fluid.layers.data(
-                    name='lab2', shape=[3, 4, 5, 6], dtype="int32"
+                lab2 = paddle.static.data(
+                    name='lab2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                 )
                 paddle.nn.functional.cross_entropy(
                     x2, lab2, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_cross_op.py b/python/paddle/fluid/tests/unittests/test_cross_op.py
index 6cc366b85c89fa5c057e37029c5ce6bf68791161..29bdf93cf1c7ba6ce671c910796e7c0965e36e6c 100644
--- a/python/paddle/fluid/tests/unittests/test_cross_op.py
+++ b/python/paddle/fluid/tests/unittests/test_cross_op.py
@@ -69,18 +69,18 @@ class TestCrossAPI(unittest.TestCase):
     def input_data(self):
         self.data_x = np.array(
             [[1.0, 1.0, 1.0], [2.0, 2.0, 2.0], [3.0, 3.0, 3.0]]
-        )
+        ).astype('float32')
         self.data_y = np.array(
             [[1.0, 1.0, 1.0], [1.0, 1.0, 1.0], [1.0, 1.0, 1.0]]
-        )
+        ).astype('float32')
 
     def test_cross_api(self):
         self.input_data()
 
         # case 1:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 3])
-            y = fluid.layers.data(name='y', shape=[-1, 3])
+            x = paddle.static.data(name='x', shape=[-1, 3], dtype="float32")
+            y = paddle.static.data(name='y', shape=[-1, 3], dtype="float32")
             z = paddle.cross(x, y, axis=1)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -95,8 +95,8 @@ class TestCrossAPI(unittest.TestCase):
 
         # case 2:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 3])
-            y = fluid.layers.data(name='y', shape=[-1, 3])
+            x = paddle.static.data(name='x', shape=[-1, 3], dtype="float32")
+            y = paddle.static.data(name='y', shape=[-1, 3], dtype="float32")
             z = paddle.cross(x, y)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
diff --git a/python/paddle/fluid/tests/unittests/test_data.py b/python/paddle/fluid/tests/unittests/test_data.py
index 25b2372e817c4ba8f49ce1a98a84e4ca8b4d6818..0e2223767dd924cfa9ec065ec67984c29ce399af 100644
--- a/python/paddle/fluid/tests/unittests/test_data.py
+++ b/python/paddle/fluid/tests/unittests/test_data.py
@@ -17,7 +17,6 @@ import unittest
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 
 
@@ -37,21 +36,6 @@ class TestApiDataError(unittest.TestCase):
 
             self.assertRaises(TypeError, test_shape_type)
 
-    def test_layers_data(self):
-        with program_guard(Program(), Program()):
-
-            # 1. The type of 'name' in layers.data must be str.
-            def test_name_type():
-                layers.data(name=1, shape=[2, 25], dtype="bool")
-
-            self.assertRaises(TypeError, test_name_type)
-
-            # 2. The type of 'shape' in layers.data must be list or tuple.
-            def test_shape_type():
-                layers.data(name='data1', shape=2, dtype="bool")
-
-            self.assertRaises(TypeError, test_shape_type)
-
 
 class TestApiStaticDataError(unittest.TestCase):
     def test_fluid_dtype(self):
@@ -81,16 +65,15 @@ class TestApiStaticDataError(unittest.TestCase):
 
             self.assertRaises(TypeError, test_shape_type)
 
-    def test_layers_data(self):
         with program_guard(Program(), Program()):
 
-            # 1. The type of 'name' in layers.data must be str.
+            # 1. The type of 'name' in paddle.static.data must be str.
             def test_name_type():
                 paddle.static.data(name=1, shape=[2, 25], dtype="bool")
 
             self.assertRaises(TypeError, test_name_type)
 
-            # 2. The type of 'shape' in layers.data must be list or tuple.
+            # 2. The type of 'shape' in paddle.static.data must be list or tuple.
             def test_shape_type():
                 paddle.static.data(name='data1', shape=2, dtype="bool")
 
@@ -102,9 +85,6 @@ class TestApiErrorWithDynamicMode(unittest.TestCase):
         with program_guard(Program(), Program()):
             paddle.disable_static()
             self.assertRaises(AssertionError, fluid.data, 'a', [2, 25])
-            self.assertRaises(
-                AssertionError, fluid.layers.data, 'b', shape=[2, 25]
-            )
             self.assertRaises(
                 AssertionError, paddle.static.data, 'c', shape=[2, 25]
             )
diff --git a/python/paddle/fluid/tests/unittests/test_data_norm_op.py b/python/paddle/fluid/tests/unittests/test_data_norm_op.py
index 2b84f2b5685af7280970105f91d91159689ff726..37fd9902736ec93997f99d16ff7bbe3a7f175434 100644
--- a/python/paddle/fluid/tests/unittests/test_data_norm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_data_norm_op.py
@@ -19,7 +19,6 @@ import numpy as np
 from op_test import OpTest
 
 import paddle
-import paddle.fluid as fluid
 import paddle.fluid.core as core
 from paddle.fluid import Program, program_guard
 from paddle.fluid.op import Operator
@@ -518,7 +517,7 @@ class TestDataNormOpWithSlotDim(OpTest):
 class TestDataNormOpErrorr(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
-            x2 = fluid.layers.data(name='x2', shape=[3, 4], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 3, 4], dtype="int32")
             # self.assertRaises(TypeError, fluid.data_norm, x2)
             paddle.static.nn.data_norm(
                 input=x2, param_attr={}, enable_scale_and_shift=True
diff --git a/python/paddle/fluid/tests/unittests/test_dataset.py b/python/paddle/fluid/tests/unittests/test_dataset.py
index fb8c9ff6e556694f4a325710d55324016d072476..f98193ea64b1b4ef909c00ae3e79284241c82ec7 100644
--- a/python/paddle/fluid/tests/unittests/test_dataset.py
+++ b/python/paddle/fluid/tests/unittests/test_dataset.py
@@ -100,8 +100,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
@@ -192,8 +192,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
@@ -257,8 +257,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
@@ -350,13 +350,13 @@ class TestDataset(unittest.TestCase):
         startup_program = fluid.Program()
         with fluid.program_guard(train_program, startup_program):
             for slot in slots[:2]:
-                var = fluid.layers.data(
-                    name=slot, shape=[1], dtype="int64", lod_level=1
+                var = paddle.static.data(
+                    name=slot, shape=[-1, 1], dtype="int64", lod_level=1
                 )
                 slots_vars.append(var)
             for slot in slots[2:]:
-                var = fluid.layers.data(
-                    name=slot, shape=[1], dtype="float32", lod_level=1
+                var = paddle.static.data(
+                    name=slot, shape=[-1, 1], dtype="float32", lod_level=1
                 )
                 slots_vars.append(var)
 
@@ -425,17 +425,17 @@ class TestDataset(unittest.TestCase):
         train_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(train_program, startup_program):
-            var1 = fluid.layers.data(
-                name="slot1", shape=[1], dtype="int64", lod_level=0
+            var1 = paddle.static.data(
+                name="slot1", shape=[-1, 1], dtype="int64", lod_level=0
             )
-            var2 = fluid.layers.data(
-                name="slot2", shape=[1], dtype="int64", lod_level=0
+            var2 = paddle.static.data(
+                name="slot2", shape=[-1, 1], dtype="int64", lod_level=0
             )
-            var3 = fluid.layers.data(
-                name="slot3", shape=[1], dtype="float32", lod_level=0
+            var3 = paddle.static.data(
+                name="slot3", shape=[-1, 1], dtype="float32", lod_level=0
             )
-            var4 = fluid.layers.data(
-                name="slot4", shape=[1], dtype="float32", lod_level=0
+            var4 = paddle.static.data(
+                name="slot4", shape=[-1, 1], dtype="float32", lod_level=0
             )
             slots_vars = [var1, var2, var3, var4]
 
@@ -498,8 +498,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1_f", "slot2_f", "slot3_f", "slot4_f"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="float32", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="float32", lod_level=1
             )
             slots_vars.append(var)
 
@@ -614,8 +614,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
@@ -682,8 +682,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1_f", "slot2_f", "slot3_f", "slot4_f"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="float32", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="float32", lod_level=1
             )
             slots_vars.append(var)
 
@@ -807,8 +807,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
@@ -872,8 +872,8 @@ class TestDataset(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
@@ -944,8 +944,8 @@ class TestDatasetWithFetchHandler(unittest.TestCase):
         slots_vars = []
         poolings = []
         for slot in slots:
-            data = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             var = fluid.layers.cast(x=data, dtype='float32')
             pool = fluid.layers.sequence_pool(input=var, pool_type='AVERAGE')
@@ -1117,8 +1117,8 @@ class TestDataset2(unittest.TestCase):
             slots = ["slot1_ff", "slot2_ff", "slot3_ff", "slot4_ff"]
             slots_vars = []
             for slot in slots:
-                var = fluid.layers.data(
-                    name=slot, shape=[1], dtype="float32", lod_level=1
+                var = paddle.static.data(
+                    name=slot, shape=[-1, 1], dtype="float32", lod_level=1
                 )
                 slots_vars.append(var)
             fake_cost = paddle.subtract(slots_vars[0], slots_vars[-1])
@@ -1187,8 +1187,8 @@ class TestDataset2(unittest.TestCase):
             slots = ["slot1_ff", "slot2_ff", "slot3_ff", "slot4_ff"]
             slots_vars = []
             for slot in slots:
-                var = fluid.layers.data(
-                    name=slot, shape=[1], dtype="float32", lod_level=1
+                var = paddle.static.data(
+                    name=slot, shape=[-1, 1], dtype="float32", lod_level=1
                 )
                 slots_vars.append(var)
             fake_cost = paddle.subtract(slots_vars[0], slots_vars[-1])
@@ -1318,8 +1318,8 @@ class TestDataset2(unittest.TestCase):
             slots = ["slot1_ff", "slot2_ff", "slot3_ff", "slot4_ff"]
             slots_vars = []
             for slot in slots:
-                var = fluid.layers.data(
-                    name=slot, shape=[1], dtype="float32", lod_level=1
+                var = paddle.static.data(
+                    name=slot, shape=[-1, 1], dtype="float32", lod_level=1
                 )
                 slots_vars.append(var)
             fake_cost = paddle.subtract(slots_vars[0], slots_vars[-1])
diff --git a/python/paddle/fluid/tests/unittests/test_dataset_consistency_inspection.py b/python/paddle/fluid/tests/unittests/test_dataset_consistency_inspection.py
index 63814b468245ed20807b816cec5785cc9bb60628..1a8d4de560ab79b37cc417fb83e6f980e61f77ec 100644
--- a/python/paddle/fluid/tests/unittests/test_dataset_consistency_inspection.py
+++ b/python/paddle/fluid/tests/unittests/test_dataset_consistency_inspection.py
@@ -394,12 +394,11 @@ class TestDataset(unittest.TestCase):
             f.write(data)
 
         slot_data = []
-        label = fluid.layers.data(
+        label = paddle.static.data(
             name="click",
             shape=[-1, 1],
             dtype="int64",
             lod_level=0,
-            append_batch_size=False,
         )
         slot_data.append(label)
 
@@ -407,56 +406,65 @@ class TestDataset(unittest.TestCase):
         len_sparse_query = 19
         for feat_name in range(1, len_sparse_query + 1):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='int64', lod_level=1
+                paddle.static.data(
+                    name=str(feat_name),
+                    shape=[-1, 1],
+                    dtype='int64',
+                    lod_level=1,
                 )
             )
 
         # sparse_url_feat_names
         for feat_name in range(len_sparse_query + 1, len_sparse_query + 5):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='int64', lod_level=1
+                paddle.static.data(
+                    name=str(feat_name),
+                    shape=[-1, 1],
+                    dtype='int64',
+                    lod_level=1,
                 )
             )
 
         # dense_feat_names
         for feat_name in range(len_sparse_query + 5, len_sparse_query + 16):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='float32'
+                paddle.static.data(
+                    name=str(feat_name), shape=[-1, 1], dtype='float32'
                 )
             )
 
         # context_feat_namess
         for feat_name in range(len_sparse_query + 16, len_sparse_query + 18):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='float32'
+                paddle.static.data(
+                    name=str(feat_name), shape=[-1, 1], dtype='float32'
                 )
             )
 
         # neg sparse_url_feat_names
         for feat_name in range(len_sparse_query + 18, len_sparse_query + 22):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='int64', lod_level=1
+                paddle.static.data(
+                    name=str(feat_name),
+                    shape=[-1, 1],
+                    dtype='int64',
+                    lod_level=1,
                 )
             )
 
         # neg dense_feat_names
         for feat_name in range(len_sparse_query + 22, len_sparse_query + 33):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='float32'
+                paddle.static.data(
+                    name=str(feat_name), shape=[-1, 1], dtype='float32'
                 )
             )
 
         # neg context_feat_namess
         for feat_name in range(len_sparse_query + 33, len_sparse_query + 35):
             slot_data.append(
-                fluid.layers.data(
-                    name=str(feat_name), shape=[1], dtype='float32'
+                paddle.static.data(
+                    name=str(feat_name), shape=[-1, 1], dtype='float32'
                 )
             )
 
diff --git a/python/paddle/fluid/tests/unittests/test_dataset_dataloader.py b/python/paddle/fluid/tests/unittests/test_dataset_dataloader.py
index 37443f6aa501a39adc7f8bf631cf93930e61b699..078ffb4e863b68d541022a6c2e112b711961445c 100644
--- a/python/paddle/fluid/tests/unittests/test_dataset_dataloader.py
+++ b/python/paddle/fluid/tests/unittests/test_dataset_dataloader.py
@@ -86,11 +86,11 @@ class DatasetLoaderTestBase(unittest.TestCase):
         main_prog = fluid.Program()
         startup_prog = fluid.Program()
         with fluid.program_guard(main_prog, startup_prog):
-            image = fluid.layers.data(
-                name='image', shape=IMAGE_SHAPE, dtype='float32'
+            image = paddle.static.data(
+                name='image', shape=[-1] + IMAGE_SHAPE, dtype='float32'
             )
-            label = fluid.layers.data(
-                name='label', shape=LABEL_SHAPE, dtype='int64'
+            label = paddle.static.data(
+                name='label', shape=[-1] + LABEL_SHAPE, dtype='int64'
             )
 
             simple_fc_net_with_inputs(image, label)
diff --git a/python/paddle/fluid/tests/unittests/test_decoupled_py_reader.py b/python/paddle/fluid/tests/unittests/test_decoupled_py_reader.py
index 6f417973ee228d654bbd71628e356a79c38f9f3f..e02282cb9bee18d2ef0db016bc39f0e77cd3e717 100644
--- a/python/paddle/fluid/tests/unittests/test_decoupled_py_reader.py
+++ b/python/paddle/fluid/tests/unittests/test_decoupled_py_reader.py
@@ -42,10 +42,12 @@ def simple_fc_net(places, use_legacy_py_reader, use_double_buffer):
 
     with fluid.unique_name.guard():
         with fluid.program_guard(main_prog, startup_prog):
-            image = fluid.layers.data(
-                name='image', shape=[784], dtype='float32'
+            image = paddle.static.data(
+                name='image', shape=[-1, 784], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             py_reader = fluid.io.PyReader(
                 feed_list=[image, label],
                 capacity=4,
diff --git a/python/paddle/fluid/tests/unittests/test_decoupled_py_reader_data_check.py b/python/paddle/fluid/tests/unittests/test_decoupled_py_reader_data_check.py
index 1a7118cfb8279c98b7d8913572263fc29c430e0e..53ef9b02ccb105c8fc63031731fcd168fde200b1 100644
--- a/python/paddle/fluid/tests/unittests/test_decoupled_py_reader_data_check.py
+++ b/python/paddle/fluid/tests/unittests/test_decoupled_py_reader_data_check.py
@@ -50,11 +50,11 @@ class TestClass(unittest.TestCase):
             main_prog = fluid.Program()
             startup_prog = fluid.Program()
             with fluid.program_guard(main_prog, startup_prog):
-                img = fluid.layers.data(
-                    shape=img_shape, dtype='float32', name='image'
+                img = paddle.static.data(
+                    shape=[-1] + img_shape, dtype='float32', name='image'
                 )
-                label = fluid.layers.data(
-                    shape=label_shape, dtype='int64', name='label'
+                label = paddle.static.data(
+                    shape=[-1] + label_shape, dtype='int64', name='label'
                 )
 
                 feeder = fluid.DataFeeder(feed_list=[img, label], place=p)
diff --git a/python/paddle/fluid/tests/unittests/test_desc_clone.py b/python/paddle/fluid/tests/unittests/test_desc_clone.py
index ed1f9a9aaf9d95ac60ed3e89ce7123e9b7be0da1..ecb49c3172fb92a9bccb420030a51febe44b48e7 100644
--- a/python/paddle/fluid/tests/unittests/test_desc_clone.py
+++ b/python/paddle/fluid/tests/unittests/test_desc_clone.py
@@ -68,8 +68,10 @@ def cnn_model(data):
 
 def get_model(batch_size):
     # Input data
-    images = fluid.layers.data(name='pixel', shape=[1, 28, 28], dtype=DTYPE)
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    images = paddle.static.data(
+        name='pixel', shape=[-1, 1, 28, 28], dtype=DTYPE
+    )
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     # Train program
     predict = cnn_model(images)
@@ -186,7 +188,7 @@ class TestCloneWithStopGradient(unittest.TestCase):
         train_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(train_program, startup_program):
-            img = fluid.layers.data(name='image', shape=[784])
+            img = paddle.static.data(name='image', shape=[-1, 784])
             hidden1 = paddle.static.nn.fc(x=img, size=200, activation='relu')
             hidden1.stop_gradient = True
             hidden2 = paddle.nn.functional.dropout(hidden1, p=0.5)
@@ -194,7 +196,9 @@ class TestCloneWithStopGradient(unittest.TestCase):
                 input=paddle.static.nn.fc(
                     hidden2, size=10, activation='softmax'
                 ),
-                label=fluid.layers.data(name='label', shape=[1], dtype='int64'),
+                label=paddle.static.data(
+                    name='label', shape=[-1, 1], dtype='int64'
+                ),
                 reduction='none',
                 use_softmax=False,
             )
@@ -214,7 +218,7 @@ class TestCloneWithStopGradientInSubBlock(unittest.TestCase):
         train_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(train_program, startup_program):
-            img = fluid.layers.data(name='image', shape=[784])
+            img = paddle.static.data(name='image', shape=[-1, 784])
             true = paddle.ones(shape=[1], dtype="float32")
             hidden1 = paddle.static.nn.fc(x=img, size=200, activation='relu')
             hidden1.stop_gradient = True
@@ -236,7 +240,9 @@ class TestCloneWithStopGradientInSubBlock(unittest.TestCase):
                 input=paddle.static.nn.fc(
                     hidden2, size=10, activation='softmax'
                 ),
-                label=fluid.layers.data(name='label', shape=[1], dtype='int64'),
+                label=paddle.static.data(
+                    name='label', shape=[-1, 1], dtype='int64'
+                ),
                 reduction='none',
                 use_softmax=False,
             )
@@ -259,7 +265,7 @@ class TestCloneWithRaise(unittest.TestCase):
         train_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(train_program, startup_program):
-            img = fluid.layers.data(name='image', shape=[784])
+            img = paddle.static.data(name='image', shape=[-1, 784])
             true = paddle.ones(shape=[1], dtype="float32")
             hidden1 = paddle.static.nn.fc(x=img, size=200, activation='relu')
             hidden1.stop_gradient = True
@@ -280,7 +286,9 @@ class TestCloneWithRaise(unittest.TestCase):
                 input=paddle.static.nn.fc(
                     hidden2, size=10, activation='softmax'
                 ),
-                label=fluid.layers.data(name='label', shape=[1], dtype='int64'),
+                label=paddle.static.data(
+                    name='label', shape=[-1, 1], dtype='int64'
+                ),
                 reduction='none',
                 use_softmax=False,
             )
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_async.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_async.py
index c80e1a68fddaddf489f8ee607caad17c85ca72e8..cb58970e88c739d873bde5d6d74a98dfe98bb2c8 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_async.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_async.py
@@ -45,8 +45,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
 
-        x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
@@ -83,8 +83,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
 
-        x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto.py
index 080319304186f752cfb9cd98740d96caee41afac..d0521a59183a1e369855ea36004cfa9fb450df78 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto.py
@@ -44,10 +44,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
         paddle.fluid.framework.switch_startup_program(startup_program)
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_async.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_async.py
index 725c2559db051bb9f79e51cb9123f27be060aa8e..c560dfa8dbb0b0d99bebcbb3f2098faa76e19ba6 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_async.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_async.py
@@ -46,12 +46,11 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
         paddle.fluid.framework.switch_startup_program(startup_program)
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
-        input_x = paddle.fluid.layers.data(
+        input_x = paddle.static.data(
             name="x",
             shape=[-1, 1],
             dtype="int64",
             lod_level=1,
-            append_batch_size=False,
         )
         x_embedding = paddle.fluid.layers.embedding(
             is_distributed=False,
@@ -63,7 +62,7 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
             ),
             is_sparse=True,
         )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=x_embedding, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_geo.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_geo.py
index c25e60793fe03639dbdcd0cdd1e124071f52b811..9eac239742743d9125f8dfca51d6d49dc8d9ca7b 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_geo.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_auto_geo.py
@@ -46,8 +46,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
 
-        input_x = paddle.fluid.layers.data(name="x", shape=[1], dtype='int64')
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 1], dtype='int64')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         emb = paddle.fluid.layers.embedding(
             input=input_x, size=[100, 10], is_sparse=True
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_geo.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_geo.py
index 3832fd5de23f72d4e83ed65ae0296db9b6a05841..c8470f64ebbc4f2a734e081604f17e994468233c 100755
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_geo.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_geo.py
@@ -43,10 +43,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
         paddle.fluid.framework.switch_startup_program(startup_program)
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
@@ -76,10 +74,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
         paddle.fluid.framework.switch_startup_program(startup_program)
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_sync.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_sync.py
index fdaa0a69c8d20839e5004cf696a037f527e4d613..f1ff43911261e6fe2148b733fcc2951b8be49acb 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_sync.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_a_sync_optimizer_sync.py
@@ -38,8 +38,8 @@ class TestFleetGradientMergeMetaOptimizer(unittest.TestCase):
     def test_gradient_merge_optimizer(self):
         fleet.init(role_maker.PaddleCloudRoleMaker())
 
-        x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_heter_program.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_heter_program.py
index 376a0d087e6d72b61309d0fdf3fb797707aed978..bc17b0d67f9908185921b996d5a793680044baeb 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_heter_program.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_heter_program.py
@@ -65,18 +65,18 @@ class TestDistFleetHeterProgram(unittest.TestCase):
         return self.strategy
 
     def build_input(self):
-        dense_input = fluid.layers.data(
-            name="dense_input", shape=[10], dtype="float32"
+        dense_input = paddle.static.data(
+            name="dense_input", shape=[-1, 10], dtype="float32"
         )
 
         sparse_input_ids = [
-            fluid.layers.data(
-                name="C" + str(i), shape=[1], lod_level=1, dtype="int64"
+            paddle.static.data(
+                name="C" + str(i), shape=[-1, 1], lod_level=1, dtype="int64"
             )
             for i in range(1, 27)
         ]
 
-        label = fluid.layers.data(name="label", shape=[1], dtype="float32")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="float32")
 
         inputs = [dense_input] + sparse_input_ids + [label]
         return inputs
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_minimize.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_minimize.py
index 472f8a6ced6fdc70df933ec086276d1a02aa3dec..b60ff0db63e7dd5c67b23fb3f18d3c2564e9d6a0 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_minimize.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_minimize.py
@@ -69,7 +69,9 @@ class TestPSMinimize(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(name="1", shape=[1], dtype="int64", lod_level=1)
+        q = paddle.static.data(
+            name="1", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
             input=q,
@@ -95,9 +97,11 @@ class TestPSMinimize(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(name="2", shape=[1], dtype="int64", lod_level=1)
+        pt = paddle.static.data(
+            name="2", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
             input=pt,
@@ -124,7 +128,9 @@ class TestPSMinimize(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(name="3", shape=[1], dtype="int64", lod_level=1)
+        nt = paddle.static.data(
+            name="3", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
             input=nt,
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps.py
index c879875f6f771524573b208625a21595b49b3dc7..a330b45b52228c8bc5ad2ddeeee7187a1705d2d1 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps.py
@@ -69,8 +69,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.layers.embedding(
@@ -99,10 +99,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.layers.embedding(
@@ -132,8 +132,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.layers.embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps11.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps11.py
index 668b64d19390cf153b8d02618817cbbf7bb9a9df..2143dc94d39e043e1a282fc679c794a231e60a01 100755
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps11.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps11.py
@@ -69,7 +69,9 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(name="1", shape=[1], dtype="int64", lod_level=1)
+        q = paddle.static.data(
+            name="1", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
             input=q,
@@ -95,9 +97,11 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(name="2", shape=[1], dtype="int64", lod_level=1)
+        pt = paddle.static.data(
+            name="2", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
             input=pt,
@@ -124,7 +128,9 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(name="3", shape=[1], dtype="int64", lod_level=1)
+        nt = paddle.static.data(
+            name="3", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
             input=nt,
@@ -204,8 +210,8 @@ class TestPSPassWithBow(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps12.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps12.py
index 5aa14fba6a5dab4266d669f1b45a8348d63e7ec3..bee3cd9eb2239e48d6df210eb1511445d25fbf1c 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps12.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps12.py
@@ -72,7 +72,9 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(name="1", shape=[1], dtype="int64", lod_level=1)
+        q = paddle.static.data(
+            name="1", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
             input=q,
@@ -98,9 +100,11 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(name="2", shape=[1], dtype="int64", lod_level=1)
+        pt = paddle.static.data(
+            name="2", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
             input=pt,
@@ -127,7 +131,9 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(name="3", shape=[1], dtype="int64", lod_level=1)
+        nt = paddle.static.data(
+            name="3", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
             input=nt,
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps13.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps13.py
index 8ecb4e2a2ad8f61f6a988e3cd516bfdd79b8f4c1..58248d325b1452e0525f68f20276017e7ad7e814 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps13.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps13.py
@@ -73,8 +73,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
@@ -101,10 +101,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
@@ -132,8 +132,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps2.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps2.py
index d8dfcda35e235dddfd64855af3029ab0eb39a299..e207fb859de54a49a39977e6fdecad9938282a60 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps2.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps2.py
@@ -72,8 +72,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
@@ -101,10 +101,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
@@ -132,8 +132,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps3.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps3.py
index c4517cc28471f2f6fd89bfea78a9fd8ad0e79751..4093fc34cc998417fd0a187a43f2f42a84777791 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps3.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps3.py
@@ -69,8 +69,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = False
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.layers.embedding(
@@ -99,10 +99,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.layers.embedding(
@@ -132,8 +132,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.layers.embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps4.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps4.py
index 93c2d48f89777a83ff7453396fe6cbfb9b538809..025b3e90b37d46a0f42eccafd8d6700a616d788c 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps4.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps4.py
@@ -69,8 +69,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
@@ -97,10 +97,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
@@ -128,8 +128,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps5.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps5.py
index ca69a778aac0395add998e54a55081eb32d04354..51bf54b3241b488549012ef82626c5fc707e4ba9 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps5.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps5.py
@@ -69,8 +69,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.layers.embedding(
@@ -99,10 +99,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.layers.embedding(
@@ -132,8 +132,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.layers.embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps6.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps6.py
index ab6bb7198c655c7eac2e71db1d13d13c7ff59fed..165a8b6240aafac399dee3ad1803bc75b82b385c 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_ps6.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_ps6.py
@@ -69,8 +69,8 @@ class TestPSPassWithBow(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(
-            name="query_ids", shape=[1], dtype="int64", lod_level=1
+        q = paddle.static.data(
+            name="query_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
@@ -97,10 +97,10 @@ class TestPSPassWithBow(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(
-            name="pos_title_ids", shape=[1], dtype="int64", lod_level=1
+        pt = paddle.static.data(
+            name="pos_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
@@ -128,8 +128,8 @@ class TestPSPassWithBow(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(
-            name="neg_title_ids", shape=[1], dtype="int64", lod_level=1
+        nt = paddle.static.data(
+            name="neg_title_ids", shape=[-1, 1], dtype="int64", lod_level=1
         )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_sparse_embedding_ctr.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_sparse_embedding_ctr.py
index 57f4615f7c9dbaecbe826f53ff23fca30c5f684d..517232fa54eb84647b01cd5d0e42822d5d24768a 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_sparse_embedding_ctr.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_sparse_embedding_ctr.py
@@ -189,26 +189,23 @@ class TestDistMnistAsync2x2WithGauss(TestFleetBase):
             """
             dnn_input_dim, lr_input_dim = 10, 10
 
-            dnn_data = fluid.layers.data(
+            dnn_data = paddle.static.data(
                 name="dnn_data",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
-            lr_data = fluid.layers.data(
+            lr_data = paddle.static.data(
                 name="lr_data",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
-            label = fluid.layers.data(
+            label = paddle.static.data(
                 name="click",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=0,
-                append_batch_size=False,
             )
 
             datas = [dnn_data, lr_data, label]
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_spmt.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_spmt.py
index 72068108d2206478d19f06516db4bcec1714a7e3..ba6e67a035095e08c0430a04c4e6e149bdc42511 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_spmt.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_spmt.py
@@ -67,7 +67,9 @@ class TestSPMT(unittest.TestCase):
         is_sparse = True
 
         # query
-        q = fluid.layers.data(name="1", shape=[1], dtype="int64", lod_level=1)
+        q = paddle.static.data(
+            name="1", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         q_emb = fluid.contrib.layers.sparse_embedding(
             input=q,
@@ -93,9 +95,11 @@ class TestSPMT(unittest.TestCase):
             ),
         )
         # label data
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         # pt
-        pt = fluid.layers.data(name="2", shape=[1], dtype="int64", lod_level=1)
+        pt = paddle.static.data(
+            name="2", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         pt_emb = fluid.contrib.layers.sparse_embedding(
             input=pt,
@@ -122,7 +126,9 @@ class TestSPMT(unittest.TestCase):
             bias_attr=fluid.ParamAttr(name="__fc_b__"),
         )
         # nt
-        nt = fluid.layers.data(name="3", shape=[1], dtype="int64", lod_level=1)
+        nt = paddle.static.data(
+            name="3", shape=[-1, 1], dtype="int64", lod_level=1
+        )
         # embedding
         nt_emb = fluid.contrib.layers.sparse_embedding(
             input=nt,
diff --git a/python/paddle/fluid/tests/unittests/test_dist_fleet_trainer_desc_config.py b/python/paddle/fluid/tests/unittests/test_dist_fleet_trainer_desc_config.py
index b13e2b8171c8cf92d9e11e12bfbbd085cd7cf9c6..e954b56d7f972e6b75d7e9c307cb0484e25a384d 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_fleet_trainer_desc_config.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_fleet_trainer_desc_config.py
@@ -39,8 +39,8 @@ class TestDistStrategyTrainerDescConfig(unittest.TestCase):
 
         fleet.init(role_maker.PaddleCloudRoleMaker())
 
-        x = paddle.fluid.layers.data(name='x', shape=[1], dtype='float32')
-        y = paddle.fluid.layers.data(name='y', shape=[1], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=x, label=y)
         avg_cost = paddle.mean(cost)
 
diff --git a/python/paddle/fluid/tests/unittests/test_dist_mnist_fleetapi.py b/python/paddle/fluid/tests/unittests/test_dist_mnist_fleetapi.py
index 0749139be840f3dc9151bfa3cdb7775c3589d72b..08d9c52c685e4a6accebc1481f8834d0208f386c 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_mnist_fleetapi.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_mnist_fleetapi.py
@@ -55,7 +55,7 @@ class FleetCollectiveTest(unittest.TestCase):
             # Operator "gen_nccl_id" has not been registered
             return
 
-        data = fluid.layers.data(name='X', shape=[1], dtype='float32')
+        data = paddle.static.data(name='X', shape=[-1, 1], dtype='float32')
         hidden = paddle.static.nn.fc(x=data, size=10)
         loss = paddle.mean(hidden)
 
diff --git a/python/paddle/fluid/tests/unittests/test_dist_train.py b/python/paddle/fluid/tests/unittests/test_dist_train.py
index 5fa4f87f95a3240dc92260629c7ce660cad6cd42..828b07baf7bbc9220a40a508bd5d8eb5eec7d613 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_train.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_train.py
@@ -23,7 +23,6 @@ from dist_test_utils import remove_ps_flag
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 import paddle.fluid.layers.ops as ops
 from paddle.fluid import core
 from paddle.fluid.layers.io import ListenAndServ, Recv, Send
@@ -83,11 +82,10 @@ class TestSendOp(unittest.TestCase):
                     dtype="float32",
                     shape=[32, 32],
                 )
-                x = layers.data(
+                x = paddle.static.data(
                     shape=[32, 32],
                     dtype='float32',
                     name="X",
-                    append_batch_size=False,
                 )
                 fluid.initializer.Constant(value=1.0)(x, main.global_block())
                 ops._scale(x=x, scale=10.0, out=out_var)
@@ -108,12 +106,7 @@ class TestSendOp(unittest.TestCase):
                 },
             )
 
-            x = layers.data(
-                shape=[32, 32],
-                dtype='float32',
-                name='X',
-                append_batch_size=False,
-            )
+            x = paddle.static.data(shape=[32, 32], dtype='float32', name='X')
             x.persistable = True
             fluid.initializer.Constant(value=2.3)(x, main.global_block())
 
@@ -141,12 +134,7 @@ class TestSendOp(unittest.TestCase):
     def run_local(self, place):
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = layers.data(
-                shape=[32, 32],
-                dtype='float32',
-                name='X',
-                append_batch_size=False,
-            )
+            x = paddle.static.data(shape=[32, 32], dtype='float32', name='X')
             fluid.initializer.Constant(value=2.3)(x, main.global_block())
             o = paddle.scale(x=x, scale=10.0)
         exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/test_dist_transpiler.py b/python/paddle/fluid/tests/unittests/test_dist_transpiler.py
index 71fb3f96d4cef8e03f19606322b109fe12813c9f..e9b8f773c743bc8cef82694f2e1a682df3509620 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_transpiler.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_transpiler.py
@@ -38,14 +38,14 @@ class TranspilerTest(unittest.TestCase):
         self.transpiler = None
 
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.1)
@@ -292,14 +292,14 @@ class TestNoSliceVar(TranspilerTest):
 
 class TestLRDecay(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(
@@ -338,14 +338,14 @@ class TestFakeInit(TranspilerTest):
     def net_conf(self):
         dict_size, embedding_size, neg_num = 10000, 8, 5
 
-        input_word = fluid.layers.data(
-            name="input_word", shape=[1], dtype='int64', lod_level=1
+        input_word = paddle.static.data(
+            name="input_word", shape=[-1, 1], dtype='int64', lod_level=1
         )
-        true_word = fluid.layers.data(
-            name='true_label', shape=[1], dtype='int64', lod_level=1
+        true_word = paddle.static.data(
+            name='true_label', shape=[-1, 1], dtype='int64', lod_level=1
         )
-        neg_word = fluid.layers.data(
-            name="neg_label", shape=[1], dtype='int64', lod_level=1
+        neg_word = paddle.static.data(
+            name="neg_label", shape=[-1, 1], dtype='int64', lod_level=1
         )
         inputs = [input_word, true_word, neg_word]
 
@@ -458,14 +458,14 @@ class TestFakeInit(TranspilerTest):
 
 class TestDecayedAdagrad(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         opt = fluid.optimizer.DecayedAdagrad(learning_rate=0.1)
@@ -478,14 +478,14 @@ class TestDecayedAdagrad(TranspilerTest):
 
 class TestFtrl(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         opt = fluid.optimizer.Ftrl(learning_rate=0.1)
@@ -498,14 +498,14 @@ class TestFtrl(TranspilerTest):
 
 class TestLRDecayConditional(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(
@@ -561,7 +561,7 @@ class TestLRDecayConditional(TranspilerTest):
 
 class TestL2Decay(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
@@ -570,7 +570,7 @@ class TestL2Decay(TranspilerTest):
             ),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=0.1)
@@ -599,14 +599,14 @@ class TestL2Decay(TranspilerTest):
 
 class TestL2DecayWithPiecewise(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         base_lr = 1.0
@@ -673,7 +673,7 @@ class TestL2DecayWithPiecewise(TranspilerTest):
 
 class TestEmptyPserverOptimizeBlocks(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         # only one parameter
         y_predict = paddle.static.nn.fc(
             x,
@@ -681,7 +681,7 @@ class TestEmptyPserverOptimizeBlocks(TranspilerTest):
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=False,
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         sgd_optimizer = fluid.optimizer.SGD(learning_rate=1.0)
@@ -715,14 +715,14 @@ class TestDistLookupTableBase(TranspilerTest):
             pool = fluid.layers.sequence_pool(input=emb, pool_type='average')
             return pool
 
-        title_ids = fluid.layers.data(
-            name='title_ids', shape=[1], dtype='int64', lod_level=1
+        title_ids = paddle.static.data(
+            name='title_ids', shape=[-1, 1], dtype='int64', lod_level=1
         )
-        brand_ids = fluid.layers.data(
-            name='brand_ids', shape=[1], dtype='int64', lod_level=1
+        brand_ids = paddle.static.data(
+            name='brand_ids', shape=[-1, 1], dtype='int64', lod_level=1
         )
-        profile_ids = fluid.layers.data(
-            name='brand_ids', shape=[1], dtype='int64', lod_level=1
+        profile_ids = paddle.static.data(
+            name='brand_ids', shape=[-1, 1], dtype='int64', lod_level=1
         )
         title_emb = emb_pool(title_ids, self.lookup_table_name, is_distributed)
         brand_emb = emb_pool(brand_ids, self.lookup_table_name, is_distributed)
@@ -737,7 +737,7 @@ class TestDistLookupTableBase(TranspilerTest):
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
 
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         cost = paddle.nn.functional.cross_entropy(
             input=predict, label=label, reduction='none', use_softmax=False
         )
@@ -1116,14 +1116,14 @@ class TestDistArgsInProgram(TestDistLookupTableBase):
 
 class TestRMSPropOptimizer(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         optimizer = fluid.optimizer.RMSProp(learning_rate=0.1)
@@ -1148,14 +1148,14 @@ class TestRMSPropOptimizer(TranspilerTest):
 
 class TestLoadSliceVar(TranspilerTest):
     def net_conf(self):
-        x = fluid.layers.data(name='x', shape=[1000], dtype='float32')
+        x = paddle.static.data(name='x', shape=[-1, 1000], dtype='float32')
         y_predict = paddle.static.nn.fc(
             x,
             size=1000,
             weight_attr=fluid.ParamAttr(name='fc_w'),
             bias_attr=fluid.ParamAttr(name='fc_b'),
         )
-        y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
         cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
         avg_cost = paddle.mean(cost)
         optimizer = fluid.optimizer.RMSProp(learning_rate=0.1)
@@ -1315,8 +1315,10 @@ class TestRemoteNce(TestDistLookupTableBase):
         sampler = "uniform"
         nid_freq_arr = np.random.dirichlet(np.ones(20) * 1000).astype('float32')
 
-        input = fluid.layers.data(name="input", shape=[10], dtype="float32")
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        input = paddle.static.data(
+            name="input", shape=[-1, 10], dtype="float32"
+        )
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
 
         w_param = (
             fluid.default_main_program()
@@ -1388,13 +1390,13 @@ class TestRemoteHsigmoid(TestDistLookupTableBase):
 
         num_total_classes = 3
 
-        input = fluid.layers.data(name="input", shape=[1], dtype="float32")
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
-        path_table = fluid.layers.data(
-            name='path_table', shape=[3], dtype='int64'
+        input = paddle.static.data(name="input", shape=[-1, 1], dtype="float32")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
+        path_table = paddle.static.data(
+            name='path_table', shape=[-1, 3], dtype='int64'
         )
-        path_code = fluid.layers.data(
-            name='path_code', shape=[3], dtype='int64'
+        path_code = paddle.static.data(
+            name='path_code', shape=[-1, 3], dtype='int64'
         )
         w_param = (
             fluid.default_main_program()
diff --git a/python/paddle/fluid/tests/unittests/test_dist_tree_index.py b/python/paddle/fluid/tests/unittests/test_dist_tree_index.py
index d5abb734733456b72e27d32dfa31bbef42a96900..b336de40cb02296de94e88c650daccef277a6b79 100644
--- a/python/paddle/fluid/tests/unittests/test_dist_tree_index.py
+++ b/python/paddle/fluid/tests/unittests/test_dist_tree_index.py
@@ -17,7 +17,6 @@ import tempfile
 import unittest
 
 import paddle
-import paddle.fluid as fluid
 from paddle.dataset.common import download
 from paddle.distributed.fleet.dataset import TreeIndex
 
@@ -25,19 +24,19 @@ paddle.enable_static()
 
 
 def create_feeds():
-    user_input = fluid.layers.data(
-        name="item_id", shape=[1], dtype="int64", lod_level=1
+    user_input = paddle.static.data(
+        name="item_id", shape=[-1, 1], dtype="int64", lod_level=1
     )
 
-    item = fluid.layers.data(
-        name="unit_id", shape=[1], dtype="int64", lod_level=1
+    item = paddle.static.data(
+        name="unit_id", shape=[-1, 1], dtype="int64", lod_level=1
     )
 
-    label = fluid.layers.data(
-        name="label", shape=[1], dtype="int64", lod_level=1
+    label = paddle.static.data(
+        name="label", shape=[-1, 1], dtype="int64", lod_level=1
     )
-    labels = fluid.layers.data(
-        name="labels", shape=[1], dtype="int64", lod_level=1
+    labels = paddle.static.data(
+        name="labels", shape=[-1, 1], dtype="int64", lod_level=1
     )
 
     feed_list = [user_input, item, label, labels]
@@ -140,7 +139,7 @@ class TestIndexSampler(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(name=slot, shape=[1], dtype="int64")
+            var = paddle.static.data(name=slot, shape=[-1, 1], dtype="int64")
             slots_vars.append(var)
 
         dataset = paddle.distributed.InMemoryDataset()
diff --git a/python/paddle/fluid/tests/unittests/test_dot_op.py b/python/paddle/fluid/tests/unittests/test_dot_op.py
index 55460c2f14b9d6dc10a802adedf8b70d4b8dcbb4..d32057bfb0d2dd893722422775b64bbe6670b50d 100644
--- a/python/paddle/fluid/tests/unittests/test_dot_op.py
+++ b/python/paddle/fluid/tests/unittests/test_dot_op.py
@@ -116,16 +116,22 @@ class TestDotOpError(unittest.TestCase):
 
             # the input dtype of elementwise_mul must be float16 or float32 or float64 or int32 or int64
             # float16 only can be set on GPU place
-            x1 = fluid.layers.data(name='x1', shape=[120], dtype="uint8")
-            y1 = fluid.layers.data(name='y1', shape=[120], dtype="uint8")
+            x1 = paddle.static.data(name='x1', shape=[-1, 120], dtype="uint8")
+            y1 = paddle.static.data(name='y1', shape=[-1, 120], dtype="uint8")
             self.assertRaises(Exception, paddle.dot, x1, y1)
 
-            x2 = fluid.layers.data(name='x2', shape=[2, 3], dtype="float32")
-            y2 = fluid.layers.data(name='y2', shape=[2, 3], dtype="float32")
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 2, 3], dtype="float32"
+            )
+            y2 = paddle.static.data(
+                name='y2', shape=[-1, 2, 3], dtype="float32"
+            )
             self.assertRaises(Exception, paddle.dot, x2, y2)
 
-            x3 = fluid.layers.data(name='x3', shape=[3], dtype="float32")
-            y3 = fluid.layers.data(name='y3', shape=[2, 3], dtype="float32")
+            x3 = paddle.static.data(name='x3', shape=[-1, 3], dtype="float32")
+            y3 = paddle.static.data(
+                name='y3', shape=[-1, 2, 3], dtype="float32"
+            )
             self.assertRaises(Exception, paddle.dot, x2, y3)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_downpoursgd.py b/python/paddle/fluid/tests/unittests/test_downpoursgd.py
index 29e022c4ff3c6f9afc494b9c145f08b246b443aa..652660b8e00d00b585fbee121d971312dbb43a29 100644
--- a/python/paddle/fluid/tests/unittests/test_downpoursgd.py
+++ b/python/paddle/fluid/tests/unittests/test_downpoursgd.py
@@ -52,12 +52,12 @@ class TestListenAndServOp(unittest.TestCase):
                     cache_path
                 )
                 os.system(cmd)
-            x = fluid.layers.data(name='x', shape=[1], dtype='int64')
+            x = paddle.static.data(name='x', shape=[-1, 1], dtype='int64')
             x_emb = fluid.layers.embedding(
                 input=x, size=[1, 2], is_distributed=True
             )
             y_predict = paddle.static.nn.fc(x=x_emb, size=1)
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
             )
@@ -116,12 +116,12 @@ class TestListenAndServOp(unittest.TestCase):
                     cache_path
                 )
                 os.system(cmd)
-            x = fluid.layers.data(name='x', shape=[1], dtype='int64')
+            x = paddle.static.data(name='x', shape=[-1, 1], dtype='int64')
             x_emb = fluid.layers.embedding(
                 input=x, size=[1, 2], is_distributed=True
             )
             y_predict = paddle.static.nn.fc(x=x_emb, size=1)
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
             )
@@ -178,12 +178,12 @@ class TestListenAndServOp(unittest.TestCase):
                     cache_path
                 )
                 os.system(cmd)
-            x = fluid.layers.data(name='x', shape=[1], dtype='int64')
+            x = paddle.static.data(name='x', shape=[-1, 1], dtype='int64')
             x_emb = fluid.layers.embedding(
                 input=x, size=[1, 2], is_distributed=True
             )
             y_predict = paddle.static.nn.fc(x=x_emb, size=1)
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
             )
diff --git a/python/paddle/fluid/tests/unittests/test_dropout_op.py b/python/paddle/fluid/tests/unittests/test_dropout_op.py
index f68b8b0561f2b4d4a7df11e4909c57b6968cf77d..9a48b877f5f22d5029cdf504b6b1077165c5946d 100644
--- a/python/paddle/fluid/tests/unittests/test_dropout_op.py
+++ b/python/paddle/fluid/tests/unittests/test_dropout_op.py
@@ -349,8 +349,8 @@ class TestDropoutOpError(unittest.TestCase):
             def test_dtype():
                 # the input dtype of dropout must be float16 or float32 or float64
                 # float16 only can be set on GPU place
-                x2 = fluid.layers.data(
-                    name='x2', shape=[3, 4, 5, 6], dtype="int32"
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                 )
                 paddle.nn.functional.dropout(x2, p=0.5)
 
diff --git a/python/paddle/fluid/tests/unittests/test_dygraph_multi_forward.py b/python/paddle/fluid/tests/unittests/test_dygraph_multi_forward.py
index e79773e8d41f46805d7b29bc09b5a8320644e42b..8e07d427e4ef158c8d7bac1604c1200bee90a6c5 100644
--- a/python/paddle/fluid/tests/unittests/test_dygraph_multi_forward.py
+++ b/python/paddle/fluid/tests/unittests/test_dygraph_multi_forward.py
@@ -164,10 +164,12 @@ class TestDygraphMultiForward(unittest.TestCase):
                 paddle.dataset.mnist.train(), batch_size=128, drop_last=True
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             cost = mnist(img)
             loss = paddle.nn.functional.cross_entropy(
                 cost, label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_eager_deletion_delete_vars.py b/python/paddle/fluid/tests/unittests/test_eager_deletion_delete_vars.py
index 43bd95270839e2a5e064229d0fba197a76c4bec5..a12a17636bfc14dd2e83f365704421112d309967 100644
--- a/python/paddle/fluid/tests/unittests/test_eager_deletion_delete_vars.py
+++ b/python/paddle/fluid/tests/unittests/test_eager_deletion_delete_vars.py
@@ -32,8 +32,8 @@ fluid.core._set_eager_deletion_mode(0.0, 1.0, True)
 
 
 def simple_fc_net():
-    image = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    image = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     hidden = image
     for _ in range(4):
         hidden = paddle.static.nn.fc(
diff --git a/python/paddle/fluid/tests/unittests/test_eager_deletion_dynamic_rnn_base.py b/python/paddle/fluid/tests/unittests/test_eager_deletion_dynamic_rnn_base.py
index f9294f152dcdedf7439c7b2220cc1c02a4c55a94..44153b6e2fe4eb10ba439ffbfe2f3600e247633b 100644
--- a/python/paddle/fluid/tests/unittests/test_eager_deletion_dynamic_rnn_base.py
+++ b/python/paddle/fluid/tests/unittests/test_eager_deletion_dynamic_rnn_base.py
@@ -41,11 +41,11 @@ def train(network, use_cuda, use_parallel_executor, batch_size=32, pass_num=2):
     reader = fake_imdb_reader(word_dict_size, batch_size * 40)
     train_reader = paddle.batch(reader, batch_size=batch_size)
 
-    data = fluid.layers.data(
-        name="words", shape=[1], dtype="int64", lod_level=1
+    data = paddle.static.data(
+        name="words", shape=[-1, 1], dtype="int64", lod_level=1
     )
 
-    label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+    label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
 
     cost = network(data, label, word_dict_size)
     cost.persistable = True
diff --git a/python/paddle/fluid/tests/unittests/test_eager_deletion_padding_rnn.py b/python/paddle/fluid/tests/unittests/test_eager_deletion_padding_rnn.py
index 80bc977f091bac9e57c5e4774e5236a96115c22c..5657eb174c30331838ece0977d6da423524a4323 100644
--- a/python/paddle/fluid/tests/unittests/test_eager_deletion_padding_rnn.py
+++ b/python/paddle/fluid/tests/unittests/test_eager_deletion_padding_rnn.py
@@ -333,30 +333,22 @@ def lm_model(
         return real_res, last_hidden, last_cell
 
     batch_size_each = batch_size
-    x = layers.data(
-        name="x",
-        shape=[batch_size_each, num_steps, 1],
-        dtype='int64',
-        append_batch_size=False,
+    x = paddle.static.data(
+        name="x", shape=[batch_size_each, num_steps, 1], dtype='int64'
     )
-    y = layers.data(
-        name="y",
-        shape=[batch_size_each * num_steps, 1],
-        dtype='int64',
-        append_batch_size=False,
+    y = paddle.static.data(
+        name="y", shape=[batch_size_each * num_steps, 1], dtype='int64'
     )
 
-    init_hidden = layers.data(
+    init_hidden = paddle.static.data(
         name="init_hidden",
         shape=[num_layers, batch_size_each, hidden_size],
         dtype='float32',
-        append_batch_size=False,
     )
-    init_cell = layers.data(
+    init_cell = paddle.static.data(
         name="init_cell",
         shape=[num_layers, batch_size_each, hidden_size],
         dtype='float32',
-        append_batch_size=False,
     )
 
     init_cell.persistable = True
diff --git a/python/paddle/fluid/tests/unittests/test_eager_deletion_recurrent_op.py b/python/paddle/fluid/tests/unittests/test_eager_deletion_recurrent_op.py
index 10f5def7248483bc019209d64f3a804bc9712d83..bd4e08819570f7838eeb51fa4861087a8d134d87 100644
--- a/python/paddle/fluid/tests/unittests/test_eager_deletion_recurrent_op.py
+++ b/python/paddle/fluid/tests/unittests/test_eager_deletion_recurrent_op.py
@@ -139,15 +139,14 @@ class EagerDeletionRecurrentOpTest1(unittest.TestCase):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot = layers.data(
-            shape=[self.input_dim], dtype='float32', name='h_boot'
+        h_boot = paddle.static.data(
+            shape=[-1, self.input_dim], dtype='float32', name='h_boot'
         )
         h_boot.stop_gradient = False
 
@@ -292,15 +291,14 @@ class EagerDeletionRecurrentOpTest2(EagerDeletionRecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot = layers.data(
-            shape=[self.input_dim], dtype='float32', name='h_boot'
+        h_boot = paddle.static.data(
+            shape=[-1, self.input_dim], dtype='float32', name='h_boot'
         )
         h_boot.stop_gradient = False
 
@@ -402,25 +400,22 @@ class EagerDeletionRecurrentOpMultipleMemoryTest(EagerDeletionRecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot1 = layers.data(
+        h_boot1 = paddle.static.data(
             shape=[self.batch_size, self.input_dim],
             dtype='float32',
             name='h_boot1',
-            append_batch_size=False,
         )
         h_boot1.stop_gradient = False
-        h_boot2 = layers.data(
+        h_boot2 = paddle.static.data(
             shape=[self.batch_size, self.input_dim],
             dtype='float32',
             name='h_boot2',
-            append_batch_size=False,
         )
         h_boot2.stop_gradient = False
 
@@ -490,11 +485,10 @@ class EagerDeletionRecurrentOpNoMemBootTest(EagerDeletionRecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
 
@@ -570,11 +564,10 @@ class EagerDeletionTwoRecurrentOpsTest(EagerDeletionRecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
 
@@ -673,15 +666,14 @@ class EagerDeletionFarwardOnlyRnnAndBackwardRnnTest(
         self.py_rnn = PySimpleRNN1(self.input_shape, self.output_shape)
 
         with fluid.program_guard(self.main_program, self.startup_program):
-            x = layers.data(
+            x = paddle.static.data(
                 shape=[self.sent_len, self.batch_size, self.input_dim],
                 dtype='float32',
                 name='x',
-                append_batch_size=False,
             )
             x.stop_gradient = False
-            h_boot = layers.data(
-                shape=[self.input_dim], dtype='float32', name='h_boot'
+            h_boot = paddle.static.data(
+                shape=[-1, self.input_dim], dtype='float32', name='h_boot'
             )
             h_boot.stop_gradient = False
 
diff --git a/python/paddle/fluid/tests/unittests/test_eager_deletion_while_op.py b/python/paddle/fluid/tests/unittests/test_eager_deletion_while_op.py
index f2b5f667fde9f40586dbb84fab79f413e365dcc3..097b5de0b90eb2b6fb54ae1da0ca58ee36624f39 100644
--- a/python/paddle/fluid/tests/unittests/test_eager_deletion_while_op.py
+++ b/python/paddle/fluid/tests/unittests/test_eager_deletion_while_op.py
@@ -66,15 +66,9 @@ class TestEagerDeletionWhileOpBase(unittest.TestCase):
                 else 1
             )
 
-        d0 = layers.data(
-            "d0", shape=[10], append_batch_size=False, dtype='float32'
-        )
-        d1 = layers.data(
-            "d1", shape=[10], append_batch_size=False, dtype='float32'
-        )
-        d2 = layers.data(
-            "d2", shape=[10], append_batch_size=False, dtype='float32'
-        )
+        d0 = paddle.static.data("d0", shape=[-1, 10], dtype='float32')
+        d1 = paddle.static.data("d1", shape=[-1, 10], dtype='float32')
+        d2 = paddle.static.data("d2", shape=[-1, 10], dtype='float32')
 
         i = layers.zeros(shape=[1], dtype='int64')
         i.stop_gradient = True
diff --git a/python/paddle/fluid/tests/unittests/test_elementwise_nn_grad.py b/python/paddle/fluid/tests/unittests/test_elementwise_nn_grad.py
index 4be4ddd2287e9bf405eec99d6355c30d9adc2739..2c5da64817d5f2fc7a54c4e121a41562ba8c4664 100644
--- a/python/paddle/fluid/tests/unittests/test_elementwise_nn_grad.py
+++ b/python/paddle/fluid/tests/unittests/test_elementwise_nn_grad.py
@@ -21,7 +21,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 
 
 class TestElementwiseMulDoubleGradCheck(unittest.TestCase):
@@ -32,8 +31,8 @@ class TestElementwiseMulDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape, dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.multiply(x, y)
@@ -61,8 +60,8 @@ class TestElementwiseMulBroadcastDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape[:-1], False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape[:-1], dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._multiply_with_axis(x, y, axis=0)
@@ -90,8 +89,8 @@ class TestElementwiseAddDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape, dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.add(x, y)
@@ -119,8 +118,8 @@ class TestElementwiseAddBroadcastDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape[:-1], False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape[:-1], dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._add_with_axis(x, y, axis=0)
@@ -151,8 +150,8 @@ class TestElementwiseSubDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape, dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.subtract(x, y)
@@ -187,8 +186,8 @@ class TestElementwiseSubBroadcastDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape[:-1], False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape[:-1], dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._subtract_with_axis(x, y, axis=0)
@@ -219,8 +218,8 @@ class TestElementwiseDivDoubleGradCheck(unittest.TestCase):
         eps = 0.0001
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape, dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._divide_with_axis(x, y, axis=0)
@@ -257,8 +256,8 @@ class TestElementwiseDivBroadcastDoubleGradCheck(unittest.TestCase):
         eps = 0.0001
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape[1:-1], False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape[1:-1], dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._divide_with_axis(x, y, axis=1)
@@ -287,8 +286,8 @@ class TestElementwiseAddTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape, dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.add(x, y)
@@ -316,8 +315,8 @@ class TestElementwiseAddBroadcastTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape[:-1], False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape[:-1], dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._add_with_axis(x, y, axis=0)
@@ -348,8 +347,8 @@ class TestElementwiseMulTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape, dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.multiply(x, y)
@@ -384,8 +383,8 @@ class TestElementwiseMulBroadcastTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
-        y = layers.data('y', shape[:-1], False, dtype)
+        x = paddle.static.data('x', shape, dtype)
+        y = paddle.static.data('y', shape[:-1], dtype)
         x.persistable = True
         y.persistable = True
         out = paddle.tensor.math._add_with_axis(x, y, axis=0)
diff --git a/python/paddle/fluid/tests/unittests/test_entry_attr.py b/python/paddle/fluid/tests/unittests/test_entry_attr.py
index 07d2ab7fa8f0ebfe2f6e262df340ca8d76d2d2da..6cc7606a7a9cecb39b9124f9cbf8af859d84d32b 100644
--- a/python/paddle/fluid/tests/unittests/test_entry_attr.py
+++ b/python/paddle/fluid/tests/unittests/test_entry_attr.py
@@ -67,12 +67,8 @@ class EntryAttrChecks(unittest.TestCase):
 
         with fluid.scope_guard(scope):
             with fluid.program_guard(prog):
-                input = fluid.layers.data(
-                    name="dnn_data",
-                    shape=[-1, 1],
-                    dtype="int64",
-                    lod_level=1,
-                    append_batch_size=False,
+                input = paddle.static.data(
+                    name="dnn_data", shape=[-1, 1], dtype="int64", lod_level=1
                 )
                 prob = ProbabilityEntry(0.5)
                 emb = paddle.static.nn.sparse_embedding(
diff --git a/python/paddle/fluid/tests/unittests/test_entry_attr2.py b/python/paddle/fluid/tests/unittests/test_entry_attr2.py
index 5db31f906fa8e306803b0f921f43a188ea291536..d06e248f901df1f8ad9493e92c97338840e769f1 100644
--- a/python/paddle/fluid/tests/unittests/test_entry_attr2.py
+++ b/python/paddle/fluid/tests/unittests/test_entry_attr2.py
@@ -28,12 +28,8 @@ class EntryAttrChecks(unittest.TestCase):
 
         with fluid.scope_guard(scope):
             with fluid.program_guard(prog):
-                input = fluid.layers.data(
-                    name="dnn_data",
-                    shape=[-1, 1],
-                    dtype="int64",
-                    lod_level=1,
-                    append_batch_size=False,
+                input = paddle.static.data(
+                    name="dnn_data", shape=[-1, 1], dtype="int64", lod_level=1
                 )
                 emb = fluid.layers.embedding(
                     input=input,
diff --git a/python/paddle/fluid/tests/unittests/test_exception.py b/python/paddle/fluid/tests/unittests/test_exception.py
index aca120b48fc7ebe712f184e77660a0a81a20d9c8..57faeaacc0708100be443e0b7e9b7f32b9848508 100644
--- a/python/paddle/fluid/tests/unittests/test_exception.py
+++ b/python/paddle/fluid/tests/unittests/test_exception.py
@@ -40,8 +40,8 @@ class TestExceptionNoCStack(unittest.TestCase):
         fluid.set_flags({'FLAGS_call_stack_level': 1})
 
     def test_exception_in_static_mode(self):
-        x = fluid.layers.data(name='X', shape=[-1, 13], dtype='float32')
-        y = fluid.layers.data(name='Y', shape=[-1, 1], dtype='float32')
+        x = paddle.static.data(name='X', shape=[-1, 13], dtype='float32')
+        y = paddle.static.data(name='Y', shape=[-1, 1], dtype='float32')
         predict = paddle.static.nn.fc(x, size=1)
         loss = paddle.nn.functional.square_error_cost(input=predict, label=y)
         avg_loss = paddle.mean(loss)
diff --git a/python/paddle/fluid/tests/unittests/test_executor_and_mul.py b/python/paddle/fluid/tests/unittests/test_executor_and_mul.py
index d1c822b5823aa6765b4dc447d68ddd05fab9d48a..f9bb4286ad38147342b0d2159f6c24b00ea82cc0 100644
--- a/python/paddle/fluid/tests/unittests/test_executor_and_mul.py
+++ b/python/paddle/fluid/tests/unittests/test_executor_and_mul.py
@@ -18,20 +18,19 @@ import numpy as np
 
 import paddle
 from paddle.fluid.executor import Executor
-from paddle.fluid.layers import data, zeros
+from paddle.fluid.layers import zeros
+from paddle.static import data
 from paddle.tensor import array_write
 
 
 class TestExecutor(unittest.TestCase):
     def test_mul(self):
         i = zeros(shape=[1], dtype='int64')
-        a = data(name='a', shape=[784], dtype='float32')
+        a = data(name='a', shape=[-1, 784], dtype='float32')
         array = array_write(x=a, i=i)
 
         i = paddle.increment(i)
-        b = data(
-            name='b', shape=[784, 100], dtype='float32', append_batch_size=False
-        )
+        b = data(name='b', shape=[784, 100], dtype='float32')
         array_write(x=b, i=i, array=array)
 
         i = paddle.increment(i)
diff --git a/python/paddle/fluid/tests/unittests/test_executor_and_use_program_cache.py b/python/paddle/fluid/tests/unittests/test_executor_and_use_program_cache.py
index 623f2065bc0e29dcf587bb5b988d97315bdf5ec9..fe9d09cb54d424e49e059f70a9122d7cd2a35343 100644
--- a/python/paddle/fluid/tests/unittests/test_executor_and_use_program_cache.py
+++ b/python/paddle/fluid/tests/unittests/test_executor_and_use_program_cache.py
@@ -26,13 +26,10 @@ class TestExecutor(unittest.TestCase):
         main_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(main_program, startup_program):
-            a = fluid.layers.data(name='a', shape=[784], dtype='float32')
-            b = fluid.layers.data(
-                name='b',
-                shape=[784, 100],
-                dtype='float32',
-                append_batch_size=False,
-            )
+            a = paddle.static.data(name='a', shape=[-1, 784], dtype='float32')
+            b = paddle.static.data(name='b', shape=[784, 100], dtype='float32')
+            a.desc.set_need_check_feed(False)
+            b.desc.set_need_check_feed(False)
             output = paddle.matmul(x=a, y=b)
 
         # Compute with numpy
diff --git a/python/paddle/fluid/tests/unittests/test_expand_as_v2_op.py b/python/paddle/fluid/tests/unittests/test_expand_as_v2_op.py
index fccfab3e4c3fd3c644efc8c82fa616fc100673a8..8fb2110bfe211f7e60ff5ca4a3c3224e5a3441c7 100755
--- a/python/paddle/fluid/tests/unittests/test_expand_as_v2_op.py
+++ b/python/paddle/fluid/tests/unittests/test_expand_as_v2_op.py
@@ -100,10 +100,10 @@ class TestExpandAsOpRank5(TestExpandAsBasic):
 class TestExpandAsV2Error(unittest.TestCase):
     def test_errors(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            x1 = fluid.layers.data(name='x1', shape=[4], dtype="uint8")
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="int32")
+            x1 = paddle.static.data(name='x1', shape=[-1, 4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="int32")
             self.assertRaises(TypeError, paddle.tensor.expand_as, x1, x2)
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tensor.expand_as, x3, x2)
 
@@ -113,14 +113,11 @@ class TestExpandAsV2API(unittest.TestCase):
     def test_api(self):
         input1 = np.random.random([12, 14]).astype("float32")
         input2 = np.random.random([2, 12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
-        )
+        x = paddle.static.data(name='x', shape=[12, 14], dtype="float32")
 
-        y = fluid.layers.data(
+        y = paddle.static.data(
             name='target_tensor',
             shape=[2, 12, 14],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/test_expand_v2_op.py b/python/paddle/fluid/tests/unittests/test_expand_v2_op.py
index 8839def6923639d8b49344f3a2f357d0179ddaf2..0a5eda417e95cd82b28f720d33729d9eb3231b43 100644
--- a/python/paddle/fluid/tests/unittests/test_expand_v2_op.py
+++ b/python/paddle/fluid/tests/unittests/test_expand_v2_op.py
@@ -21,7 +21,6 @@ from op_test import OpTest
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, core, program_guard
 
 
@@ -194,9 +193,9 @@ class TestExpandV2Error(unittest.TestCase):
             )
             shape = [2, 2]
             self.assertRaises(TypeError, paddle.tensor.expand, x1, shape)
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="uint8")
             self.assertRaises(TypeError, paddle.tensor.expand, x2, shape)
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tensor.expand, x3, shape)
 
@@ -205,15 +204,12 @@ class TestExpandV2Error(unittest.TestCase):
 class TestExpandV2API(unittest.TestCase):
     def test_api(self):
         input = np.random.random([12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
-        )
+        x = paddle.static.data(name='x', shape=[12, 14], dtype="float32")
 
         positive_2 = fluid.layers.fill_constant([1], "int32", 12)
-        expand_shape = fluid.layers.data(
+        expand_shape = paddle.static.data(
             name="expand_shape",
             shape=[2],
-            append_batch_size=False,
             dtype="int32",
         )
 
@@ -273,7 +269,7 @@ class TestExpandDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3], False, dtype)
+        data = paddle.static.data('data', [2, 3], dtype)
         data.persistable = True
         out = paddle.expand(data, [2, 3])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -304,7 +300,7 @@ class TestExpandTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3], False, dtype)
+        data = paddle.static.data('data', [2, 3], dtype)
         data.persistable = True
         out = paddle.expand(data, [2, 3])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_fc_op.py b/python/paddle/fluid/tests/unittests/test_fc_op.py
index b07fc1d0001a5ab2e1b63cbce0a9f6f2c0660ad0..03ef6c3db6c190216c7206f7e7fabe69eb2ffbf9 100644
--- a/python/paddle/fluid/tests/unittests/test_fc_op.py
+++ b/python/paddle/fluid/tests/unittests/test_fc_op.py
@@ -146,10 +146,9 @@ class TestFcOp_NumFlattenDims_NegOne(unittest.TestCase):
 
             with program_guard(main_program, startup_program):
                 input = np.random.random([2, 2, 25]).astype("float32")
-                x = fluid.layers.data(
+                x = paddle.static.data(
                     name="x",
                     shape=[2, 2, 25],
-                    append_batch_size=False,
                     dtype="float32",
                 )
 
@@ -191,13 +190,13 @@ class TestFCOpError(unittest.TestCase):
 
             def test_type():
                 # dtype must be float32 or float64
-                x2 = fluid.layers.data(name='x2', shape=[4], dtype='int32')
+                x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype='int32')
                 paddle.static.nn.fc(x=x2, size=1)
 
             self.assertRaises(TypeError, test_type)
 
             # The input dtype of fc can be float16 in GPU, test for warning
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype='float16')
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype='float16')
             paddle.static.nn.fc(x=x3, size=1)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_feed_data_check_shape_type.py b/python/paddle/fluid/tests/unittests/test_feed_data_check_shape_type.py
index b642e2524ee292f87ac24f6b9396b91b0e4f04c1..0e5330014d3f5f3891a8db32cc67e85a4e0448c3 100644
--- a/python/paddle/fluid/tests/unittests/test_feed_data_check_shape_type.py
+++ b/python/paddle/fluid/tests/unittests/test_feed_data_check_shape_type.py
@@ -30,7 +30,7 @@ np.random.seed(123)
 class TestFeedData(unittest.TestCase):
     '''
     Test paddle.fluid.data feeds with different shape and types.
-    Note: paddle.fluid.data is not paddle.fluid.layers.data.
+    Note: paddle.fluid.data is not paddle.static.data.
     '''
 
     def setUp(self):
diff --git a/python/paddle/fluid/tests/unittests/test_fetch_unmerged.py b/python/paddle/fluid/tests/unittests/test_fetch_unmerged.py
index d93ee36b6e2409d793e5a35a0d281150628146ee..f7313f93e0d14e01d29693b458b7f43b518e52cc 100644
--- a/python/paddle/fluid/tests/unittests/test_fetch_unmerged.py
+++ b/python/paddle/fluid/tests/unittests/test_fetch_unmerged.py
@@ -57,11 +57,11 @@ class TestFetchUnmerged(unittest.TestCase):
     def build_program(self, main, startup, is_test):
         with fluid.unique_name.guard():
             with fluid.program_guard(main, startup):
-                img = fluid.layers.data(
-                    name='image', shape=[1, 28, 28], dtype='float32'
+                img = paddle.static.data(
+                    name='image', shape=[-1, 1, 28, 28], dtype='float32'
                 )
-                label = fluid.layers.data(
-                    name='label', shape=[1], dtype='int64'
+                label = paddle.static.data(
+                    name='label', shape=[-1, 1], dtype='int64'
                 )
                 loss, prediction = self.conv_net(img, label)
                 if not is_test:
diff --git a/python/paddle/fluid/tests/unittests/test_fill_constant_op.py b/python/paddle/fluid/tests/unittests/test_fill_constant_op.py
index 5e1af99259db1acd868da01614399f286101db3f..38ef0379747dbaa5aa2709efbc3033b9b1966b55 100644
--- a/python/paddle/fluid/tests/unittests/test_fill_constant_op.py
+++ b/python/paddle/fluid/tests/unittests/test_fill_constant_op.py
@@ -379,7 +379,7 @@ class TestFillConstantOpError(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
             # for ci coverage
-            x1 = fluid.layers.data(name='x1', shape=[1], dtype="int16")
+            x1 = paddle.static.data(name='x1', shape=[-1, 1], dtype="int16")
             self.assertRaises(
                 TypeError,
                 fluid.layers.fill_constant,
@@ -399,7 +399,7 @@ class TestFillConstantOpError(unittest.TestCase):
 
             # The argument dtype of fill_constant_op must be one of bool, float16,
             # float32, float64, uint8, int16, int32 or int64
-            x2 = fluid.layers.data(name='x2', shape=[1], dtype="int32")
+            x2 = paddle.static.data(name='x2', shape=[-1, 1], dtype="int32")
 
             self.assertRaises(
                 TypeError,
diff --git a/python/paddle/fluid/tests/unittests/test_fleet.py b/python/paddle/fluid/tests/unittests/test_fleet.py
index bc5a083d17d4c1e92647ed4f051fd36b70c0e21e..736f68be35806fd219d23ee852f0fdac039f9856 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet.py
@@ -52,12 +52,11 @@ class TestFleet1(unittest.TestCase):
         startup_program = fluid.Program()
         scope = fluid.Scope()
         with fluid.program_guard(train_program, startup_program):
-            show = fluid.layers.data(
+            show = paddle.static.data(
                 name="show",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
             emb = fluid.layers.embedding(
                 input=show,
@@ -71,12 +70,11 @@ class TestFleet1(unittest.TestCase):
                 input=bow, epsilon=1e-4, name="norm"
             )
             fc = paddle.static.nn.fc(x=bow, size=1, activation=None)
-            label = fluid.layers.data(
+            label = paddle.static.data(
                 name="click",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
             label_cast = fluid.layers.cast(label, dtype='float32')
             cost = paddle.nn.functional.log_loss(fc, label_cast)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_api_input.py b/python/paddle/fluid/tests/unittests/test_fleet_api_input.py
index 12acfdf76321d3ab95b29c9845ae9d0f1052d2c6..b57a30d75266d3676718b93099dbffb6b61c423a 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_api_input.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_api_input.py
@@ -61,7 +61,7 @@ class FleetTest(unittest.TestCase):
         self.assertRaises(Exception, fleet.split_files, "files")
         self.assertRaises(Exception, fleet.init, "pserver")
 
-        data = fluid.layers.data(name='X', shape=[1], dtype='float32')
+        data = paddle.static.data(name='X', shape=[-1, 1], dtype='float32')
         hidden = paddle.static.nn.fc(x=data, size=10)
         loss = paddle.mean(hidden)
         adam = fluid.optimizer.Adam()
@@ -176,7 +176,7 @@ class TranspilerOptimizerTest(unittest.TestCase):
 
         transpiler = TranspilerOptimizer(fluid.optimizer.Adam(0.001))
         self.assertRaises(Exception, transpiler.minimize, loss=[])
-        data = fluid.layers.data(name='X', shape=[1], dtype='float32')
+        data = paddle.static.data(name='X', shape=[-1, 1], dtype='float32')
         hidden = paddle.static.nn.fc(x=data, size=10)
         loss = paddle.mean(hidden)
         self.assertRaises(
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_auto.py b/python/paddle/fluid/tests/unittests/test_fleet_auto.py
index 1e30f703ff893b64a15af32316f6a4caaacc9b29..b9b18f04fac6e536e61002bdffa1345d8bba3dea 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_auto.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_auto.py
@@ -32,10 +32,8 @@ class TestDistributedStrategyAuto(unittest.TestCase):
 
     def test_distributed_strategy_auto(self):
         fleet.init(is_collective=True)
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_base_2.py b/python/paddle/fluid/tests/unittests/test_fleet_base_2.py
index 4a56f8913a434a76080a0f61b633293362a687b4..ee5a84d1e41e810e2dc6a47758fd672e91223eb0 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_base_2.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_base_2.py
@@ -39,13 +39,11 @@ class TestFleetBase(unittest.TestCase):
         os.environ["TRAINING_ROLE"] = "TRAINER"
         os.environ["PADDLE_TRAINER_ID"] = "1"
 
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_slot = paddle.static.data(
+            name="slot", shape=[-1, 1], dtype='int64'
         )
-        input_slot = paddle.fluid.layers.data(
-            name="slot", shape=[1], dtype='int64'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         emb = paddle.fluid.layers.embedding(
             input=input_slot, size=[10, 9], is_sparse=True
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_base_3.py b/python/paddle/fluid/tests/unittests/test_fleet_base_3.py
index 30cdf5fbed4b7267c2d404e8d7010afff40a2f60..e24beee28e12b85e16abf1e5738f5e9b934835d5 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_base_3.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_base_3.py
@@ -32,10 +32,8 @@ class TestFleetBase_1(unittest.TestCase):
         ] = "127.0.0.1:36001,127.0.0.2:36001"
 
     def test_collective_minimize(self):
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
@@ -63,10 +61,8 @@ class TestFleetBase(unittest.TestCase):
         ] = "127.0.0.1:36001,127.0.0.2:36001"
 
     def test_fleet_get_applied_optimizer(self):
-        input_x = paddle.fluid.layers.data(
-            name="x", shape=[32], dtype='float32'
-        )
-        input_y = paddle.fluid.layers.data(name="y", shape=[1], dtype='int64')
+        input_x = paddle.static.data(name="x", shape=[-1, 32], dtype='float32')
+        input_y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
 
         fc_1 = paddle.static.nn.fc(x=input_x, size=64, activation='tanh')
         fc_2 = paddle.static.nn.fc(x=fc_1, size=64, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_executor.py b/python/paddle/fluid/tests/unittests/test_fleet_executor.py
index 400009f820de3c59cafb87582ca43c77dc7ae176..d798ffb016c536aaf763136b8a5e6dec560f162a 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_executor.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_executor.py
@@ -45,12 +45,14 @@ class TestFleetExecutor(unittest.TestCase):
         exe = paddle.static.Executor(place)
         empty_program = paddle.static.Program()
         with fluid.program_guard(empty_program, empty_program):
-            x = fluid.layers.data(
-                name='x', shape=x_data.shape, dtype=x_data.dtype
+            x = paddle.static.data(
+                name='x', shape=[-1] + list(x_data.shape), dtype=x_data.dtype
             )
-            y = fluid.layers.data(
-                name='y', shape=y_data.shape, dtype=y_data.dtype
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(
+                name='y', shape=[-1] + list(y_data.shape), dtype=y_data.dtype
             )
+            y.desc.set_need_check_feed(False)
             z = x + y
             a = 2 * x + 3 * y
             loss = paddle.mean(a)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_executor_multi_devices.py b/python/paddle/fluid/tests/unittests/test_fleet_executor_multi_devices.py
index b2a1c488dcec66016c00d3d7e8b1de9b22b91d57..53bd4a20bcd12304b61272cb4e80def3c4324136 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_executor_multi_devices.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_executor_multi_devices.py
@@ -27,7 +27,9 @@ class TestFleetExecutor(unittest.TestCase):
         exe = paddle.static.Executor(place)
         empty_program = paddle.static.Program()
         with fluid.program_guard(empty_program, empty_program):
-            x = fluid.layers.data(name='x', shape=[1], dtype=paddle.float32)
+            x = paddle.static.data(
+                name='x', shape=[-1, 1], dtype=paddle.float32
+            )
         empty_program._pipeline_opt = {
             "fleet_opt": fleet_opt,
             "section_program": empty_program,
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_executor_origin_scheduler.py b/python/paddle/fluid/tests/unittests/test_fleet_executor_origin_scheduler.py
index d24348b7d77b58234f2dbc1ef9d7ae7d563a19d3..726687d87df9188cdab1274addddef0aad29dcda 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_executor_origin_scheduler.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_executor_origin_scheduler.py
@@ -45,12 +45,14 @@ class TestFleetExecutor(unittest.TestCase):
         exe = paddle.static.Executor(place)
         empty_program = paddle.static.Program()
         with fluid.program_guard(empty_program, empty_program):
-            x = fluid.layers.data(
-                name='x', shape=x_data.shape, dtype=x_data.dtype
+            x = paddle.static.data(
+                name='x', shape=[-1] + list(x_data.shape), dtype=x_data.dtype
             )
-            y = fluid.layers.data(
-                name='y', shape=y_data.shape, dtype=y_data.dtype
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(
+                name='y', shape=[-1] + list(y_data.shape), dtype=y_data.dtype
             )
+            y.desc.set_need_check_feed(False)
             z = x + y
             a = 2 * x + 3 * y
             loss = paddle.mean(a)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_executor_with_task_nodes.py b/python/paddle/fluid/tests/unittests/test_fleet_executor_with_task_nodes.py
index 46eb0dc6f0bf8428ca0b5b6989fb6444ca5b2495..8c12fb973524672482871f6933e1590623fee431 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_executor_with_task_nodes.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_executor_with_task_nodes.py
@@ -28,12 +28,14 @@ class TestFleetExecutor(unittest.TestCase):
         exe = paddle.static.Executor(place)
         empty_program = paddle.static.Program()
         with fluid.program_guard(empty_program, empty_program):
-            x = fluid.layers.data(
-                name='x', shape=x_data.shape, dtype=x_data.dtype
+            x = paddle.static.data(
+                name='x', shape=[-1] + list(x_data.shape), dtype=x_data.dtype
             )
-            y = fluid.layers.data(
-                name='y', shape=y_data.shape, dtype=y_data.dtype
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(
+                name='y', shape=[-1] + list(y_data.shape), dtype=y_data.dtype
             )
+            y.desc.set_need_check_feed(False)
             z = x + y
             a = 2 * x + 3 * y
             loss = paddle.mean(a)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_nocvm_1.py b/python/paddle/fluid/tests/unittests/test_fleet_nocvm_1.py
index c651a456fd5112e802dfee765542a4d04737d844..09a9db8ccd5739604a1dd38a8fbad627fd6a6013 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_nocvm_1.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_nocvm_1.py
@@ -52,12 +52,11 @@ class TestFleet1(unittest.TestCase):
         startup_program = fluid.Program()
         scope = fluid.Scope()
         with fluid.program_guard(train_program, startup_program):
-            show = fluid.layers.data(
+            show = paddle.static.data(
                 name="show",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
             emb = fluid.layers.embedding(
                 input=show,
@@ -67,12 +66,11 @@ class TestFleet1(unittest.TestCase):
                 param_attr=fluid.ParamAttr(name="embedding"),
             )
             fc = paddle.static.nn.fc(x=emb, size=1, activation=None)
-            label = fluid.layers.data(
+            label = paddle.static.data(
                 name="click",
                 shape=[-1, 1],
                 dtype="int64",
                 lod_level=1,
-                append_batch_size=False,
             )
             label_cast = fluid.layers.cast(label, dtype='float32')
             cost = paddle.nn.functional.log_loss(fc, label_cast)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_rolemaker.py b/python/paddle/fluid/tests/unittests/test_fleet_rolemaker.py
index 4c3c321ac0ad3c103ef0ecb828cd4eb9483f84f5..893956587eb81fc63a30502f20cdb17adb4d2dda 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_rolemaker.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_rolemaker.py
@@ -82,20 +82,12 @@ class TestCloudRoleMaker(unittest.TestCase):
         startup_program = fluid.Program()
         scope = fluid.Scope()
         with fluid.program_guard(train_program, startup_program):
-            show = fluid.layers.data(
-                name="show",
-                shape=[-1, 1],
-                dtype="float32",
-                lod_level=1,
-                append_batch_size=False,
+            show = paddle.static.data(
+                name="show", shape=[-1, 1], dtype="float32", lod_level=1
             )
             fc = paddle.static.nn.fc(x=show, size=1, activation=None)
-            label = fluid.layers.data(
-                name="click",
-                shape=[-1, 1],
-                dtype="int64",
-                lod_level=1,
-                append_batch_size=False,
+            label = paddle.static.data(
+                name="click", shape=[-1, 1], dtype="int64", lod_level=1
             )
             label_cast = fluid.layers.cast(label, dtype='float32')
             cost = paddle.nn.functional.log_loss(fc, label_cast)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_2.py b/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_2.py
index 50a6013e2d2172b33356e6f7dc2bb0800cac2cb9..ece84ed1d5bc785c7159c4207134cabff6f22c95 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_2.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_2.py
@@ -63,20 +63,12 @@ class TestCloudRoleMaker2(unittest.TestCase):
         startup_program = fluid.Program()
         scope = fluid.Scope()
         with fluid.program_guard(train_program, startup_program):
-            show = fluid.layers.data(
-                name="show",
-                shape=[-1, 1],
-                dtype="float32",
-                lod_level=1,
-                append_batch_size=False,
+            show = paddle.static.data(
+                name="show", shape=[-1, 1], dtype="float32", lod_level=1
             )
             fc = paddle.static.nn.fc(x=show, size=1, activation=None)
-            label = fluid.layers.data(
-                name="click",
-                shape=[-1, 1],
-                dtype="int64",
-                lod_level=1,
-                append_batch_size=False,
+            label = paddle.static.data(
+                name="click", shape=[-1, 1], dtype="int64", lod_level=1
             )
             label_cast = fluid.layers.cast(label, dtype='float32')
             cost = paddle.nn.functional.log_loss(fc, label_cast)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_3.py b/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_3.py
index 27cb171c0dd6d285a4526224ae2b2210c8ff27be..94e293978b59756c516f0d6a667b3f45324b199a 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_3.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_rolemaker_3.py
@@ -56,20 +56,12 @@ class TestCloudRoleMaker(unittest.TestCase):
         startup_program = fluid.Program()
         scope = fluid.Scope()
         with fluid.program_guard(train_program, startup_program):
-            show = fluid.layers.data(
-                name="show",
-                shape=[-1, 1],
-                dtype="float32",
-                lod_level=1,
-                append_batch_size=False,
+            show = paddle.static.data(
+                name="show", shape=[-1, 1], dtype="float32", lod_level=1
             )
             fc = paddle.static.nn.fc(x=show, size=1, activation=None)
-            label = fluid.layers.data(
-                name="click",
-                shape=[-1, 1],
-                dtype="int64",
-                lod_level=1,
-                append_batch_size=False,
+            label = paddle.static.data(
+                name="click", shape=[-1, 1], dtype="int64", lod_level=1
             )
             label_cast = fluid.layers.cast(label, dtype='float32')
             cost = paddle.nn.functional.log_loss(fc, label_cast)
diff --git a/python/paddle/fluid/tests/unittests/test_fleet_unitaccessor.py b/python/paddle/fluid/tests/unittests/test_fleet_unitaccessor.py
index 178fcfa230e449c4d0a1e8c96f2a3401a728dbdc..3eb24d9b4015b6b630585b54325b1c790148ac4f 100644
--- a/python/paddle/fluid/tests/unittests/test_fleet_unitaccessor.py
+++ b/python/paddle/fluid/tests/unittests/test_fleet_unitaccessor.py
@@ -52,12 +52,8 @@ class TestFleet1(unittest.TestCase):
         startup_program = fluid.Program()
         scope = fluid.Scope()
         with fluid.program_guard(train_program, startup_program):
-            show = fluid.layers.data(
-                name="show",
-                shape=[-1, 1],
-                dtype="int64",
-                lod_level=1,
-                append_batch_size=False,
+            show = paddle.static.data(
+                name="show", shape=[-1, 1], dtype="int64", lod_level=1
             )
             emb = fluid.layers.embedding(
                 input=show,
@@ -67,12 +63,8 @@ class TestFleet1(unittest.TestCase):
                 param_attr=fluid.ParamAttr(name="embedding"),
             )
             fc = paddle.static.nn.fc(x=emb, size=1, activation=None)
-            label = fluid.layers.data(
-                name="click",
-                shape=[-1, 1],
-                dtype="int64",
-                lod_level=1,
-                append_batch_size=False,
+            label = paddle.static.data(
+                name="click", shape=[-1, 1], dtype="int64", lod_level=1
             )
             label_cast = fluid.layers.cast(label, dtype='float32')
             cost = paddle.nn.functional.log_loss(fc, label_cast)
diff --git a/python/paddle/fluid/tests/unittests/test_flip.py b/python/paddle/fluid/tests/unittests/test_flip.py
index f7ebd3d1d54f9534459b8c713a5b4231bbc7ba72..1807199821eb7459c3573877928f8f44172ba129 100644
--- a/python/paddle/fluid/tests/unittests/test_flip.py
+++ b/python/paddle/fluid/tests/unittests/test_flip.py
@@ -22,7 +22,6 @@ from op_test import OpTest
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 
 
 class TestFlipOp_API(unittest.TestCase):
@@ -145,7 +144,7 @@ class TestFlipDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [3, 2, 2], False, dtype)
+        data = paddle.static.data('data', [3, 2, 2], dtype)
         data.persistable = True
         out = paddle.flip(data, [0, 1])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -177,7 +176,7 @@ class TestFlipTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [3, 2, 2], False, dtype)
+        data = paddle.static.data('data', [3, 2, 2], dtype)
         data.persistable = True
         out = paddle.flip(data, [0, 1])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_fuse_bn_act_pass.py b/python/paddle/fluid/tests/unittests/test_fuse_bn_act_pass.py
index 9e067454482d79b8292a35fd79d357c4d0ab0f78..4d7fb60d4660e16947bb62734d66cd48d36691f0 100644
--- a/python/paddle/fluid/tests/unittests/test_fuse_bn_act_pass.py
+++ b/python/paddle/fluid/tests/unittests/test_fuse_bn_act_pass.py
@@ -21,8 +21,10 @@ import paddle.fluid as fluid
 class TestFuseBatchNormActPass(unittest.TestCase):
     def build_program(self, main_program, startup_program, use_cuda, seed=1):
         with fluid.program_guard(main_program, startup_program):
-            x = fluid.layers.data(name='x', shape=[1, 28, 28], dtype='float32')
-            y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+            x = paddle.static.data(
+                name='x', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
             hidden1 = paddle.static.nn.conv2d(
                 input=x,
                 filter_size=3,
diff --git a/python/paddle/fluid/tests/unittests/test_fuse_bn_add_act_pass.py b/python/paddle/fluid/tests/unittests/test_fuse_bn_add_act_pass.py
index a8d0f89c86649d3e823b7c1f18d348b2fa9d105b..d981ccbe14ccbafd3bc7ec1f9a54316965c37908 100644
--- a/python/paddle/fluid/tests/unittests/test_fuse_bn_add_act_pass.py
+++ b/python/paddle/fluid/tests/unittests/test_fuse_bn_add_act_pass.py
@@ -64,8 +64,10 @@ class TestFusedBnAddActAPI(unittest.TestCase):
         self, main_program, startup_program, use_cuda, seed=1
     ):
         with fluid.program_guard(main_program, startup_program):
-            x = fluid.layers.data(name='x', shape=[1, 28, 28], dtype='float32')
-            y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+            x = paddle.static.data(
+                name='x', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
             conv1_1 = paddle.static.nn.conv2d(
                 input=x,
                 filter_size=3,
@@ -123,8 +125,10 @@ class TestFusedBnAddActAPI(unittest.TestCase):
         self, main_program, startup_program, use_cuda, seed=1
     ):
         with fluid.program_guard(main_program, startup_program):
-            x = fluid.layers.data(name='x', shape=[1, 28, 28], dtype='float32')
-            y = fluid.layers.data(name="y", shape=[1], dtype='int64')
+            x = paddle.static.data(
+                name='x', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='int64')
             conv1_1 = paddle.static.nn.conv2d(
                 input=x,
                 filter_size=3,
diff --git a/python/paddle/fluid/tests/unittests/test_fuse_relu_depthwise_conv_pass.py b/python/paddle/fluid/tests/unittests/test_fuse_relu_depthwise_conv_pass.py
index 388025fb2adcb86fa4dc043843a05dca7c726065..c4f576ce2036c1e2b44ff2b4a3f66fdc19909c21 100644
--- a/python/paddle/fluid/tests/unittests/test_fuse_relu_depthwise_conv_pass.py
+++ b/python/paddle/fluid/tests/unittests/test_fuse_relu_depthwise_conv_pass.py
@@ -55,8 +55,8 @@ def sep_conv(input, channel, stride, filter, dilation=1, act=None):
 
 def simple_depthwise_net(use_feed):
     assert use_feed
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     hidden = paddle.reshape(img, (-1, 1, 28, 28))
     for _ in range(4):
         hidden = sep_conv(hidden, channel=200, stride=2, filter=5)
diff --git a/python/paddle/fluid/tests/unittests/test_fused_emb_seq_pool_op.py b/python/paddle/fluid/tests/unittests/test_fused_emb_seq_pool_op.py
index 3e3f0ca05a62f281c1f56f85ed2e044580a73f14..460bf43bbe091be66153209f10cb39010541e8db 100644
--- a/python/paddle/fluid/tests/unittests/test_fused_emb_seq_pool_op.py
+++ b/python/paddle/fluid/tests/unittests/test_fused_emb_seq_pool_op.py
@@ -18,6 +18,7 @@ import unittest
 import numpy as np
 from op_test import OpTest, skip_check_grad_ci
 
+import paddle
 import paddle.version as ver
 
 
@@ -108,8 +109,8 @@ class TestFusedEmbeddingSeqPoolApi(unittest.TestCase):
             import paddle.fluid as fluid
 
             dict_size = 20
-            data_t = fluid.layers.data(
-                name='word', shape=[1], dtype='int64', lod_level=1
+            data_t = paddle.static.data(
+                name='word', shape=[-1, 1], dtype='int64', lod_level=1
             )
             padding_idx = np.random.randint(1, 10)
             out = fluid.contrib.fused_embedding_seq_pool(
diff --git a/python/paddle/fluid/tests/unittests/test_gather_nd_op.py b/python/paddle/fluid/tests/unittests/test_gather_nd_op.py
index 6a3f5cf8556809bbecbda4cc18bd50d56e49ccb4..6c2b6d15f6f639b00b914b1bc087cc88bc750b25 100644
--- a/python/paddle/fluid/tests/unittests/test_gather_nd_op.py
+++ b/python/paddle/fluid/tests/unittests/test_gather_nd_op.py
@@ -160,20 +160,28 @@ class TestGatherNdOpWithHighRankDiff(OpTest):
 # Test Python API
 class TestGatherNdOpAPI(unittest.TestCase):
     def test_case1(self):
-        x1 = fluid.layers.data(
-            name='x1', shape=[30, 40, 50, 60], dtype='float32'
+        x1 = paddle.static.data(
+            name='x1', shape=[-1, 30, 40, 50, 60], dtype='float32'
+        )
+        index1 = paddle.static.data(
+            name='index1', shape=[-1, 2, 4], dtype='int32'
         )
-        index1 = fluid.layers.data(name='index1', shape=[2, 4], dtype='int32')
         output1 = paddle.gather_nd(x1, index1)
 
     def test_case2(self):
-        x2 = fluid.layers.data(name='x2', shape=[30, 40, 50], dtype='float32')
-        index2 = fluid.layers.data(name='index2', shape=[2, 2], dtype='int64')
+        x2 = paddle.static.data(
+            name='x2', shape=[-1, 30, 40, 50], dtype='float32'
+        )
+        index2 = paddle.static.data(
+            name='index2', shape=[-1, 2, 2], dtype='int64'
+        )
         output2 = paddle.gather_nd(x2, index2)
 
     def test_case3(self):
-        x3 = fluid.layers.data(name='x3', shape=[3, 4, 5], dtype='float32')
-        index3 = fluid.layers.data(name='index3', shape=[2, 1], dtype='int32')
+        x3 = paddle.static.data(name='x3', shape=[-1, 3, 4, 5], dtype='float32')
+        index3 = paddle.static.data(
+            name='index3', shape=[-1, 2, 1], dtype='int32'
+        )
         output3 = paddle.gather_nd(x3, index3, name="gather_nd_layer")
 
 
@@ -182,11 +190,11 @@ class TestGatherNdOpRaise(unittest.TestCase):
     def test_check_raise(self):
         def check_raise_is_test():
             try:
-                x = fluid.layers.data(
-                    name='x', shape=[3, 4, 5], dtype='float32'
+                x = paddle.static.data(
+                    name='x', shape=[-1, 3, 4, 5], dtype='float32'
                 )
-                index = fluid.layers.data(
-                    name='index', shape=[2, 10], dtype='int32'
+                index = paddle.static.data(
+                    name='index', shape=[-1, 2, 10], dtype='int32'
                 )
                 output = paddle.gather_nd(x, index)
             except Exception as e:
@@ -231,13 +239,15 @@ class TestGatherNdError(unittest.TestCase):
 class TestGatherNdAPI2(unittest.TestCase):
     def test_static(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[-1, 2], dtype='float64')
-            index = fluid.layers.data('index', shape=[-1, 1], dtype='int32')
+            data1 = paddle.static.data('data1', shape=[-1, 2], dtype='float64')
+            data1.desc.set_need_check_feed(False)
+            index = paddle.static.data('index', shape=[-1, 1], dtype='int32')
+            index.desc.set_need_check_feed(False)
             out = paddle.gather_nd(data1, index)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
             input = np.array([[1, 2], [3, 4], [5, 6]])
-            index_1 = np.array([[1]])
+            index_1 = np.array([[1]]).astype('int32')
             (result,) = exe.run(
                 feed={"data1": input, "index": index_1}, fetch_list=[out]
             )
diff --git a/python/paddle/fluid/tests/unittests/test_gather_op.py b/python/paddle/fluid/tests/unittests/test_gather_op.py
index 44ab250c7652c223affcf5f3f15190d73256a2e6..2f2538769a3b38e5a750d0dabfa035c13fa89a8a 100644
--- a/python/paddle/fluid/tests/unittests/test_gather_op.py
+++ b/python/paddle/fluid/tests/unittests/test_gather_op.py
@@ -225,8 +225,10 @@ class TestGatherOp4(TestGatherOp1):
 class API_TestGather(unittest.TestCase):
     def test_out1(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[-1, 2], dtype='float64')
-            index = fluid.layers.data('index', shape=[-1, 1], dtype='int32')
+            data1 = paddle.static.data('data1', shape=[-1, 2], dtype='float64')
+            data1.desc.set_need_check_feed(False)
+            index = paddle.static.data('index', shape=[-1, 1], dtype='int32')
+            index.desc.set_need_check_feed(False)
             out = paddle.gather(data1, index)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/test_gather_tree_op.py b/python/paddle/fluid/tests/unittests/test_gather_tree_op.py
index bcd319ed2d24a22ef86169b457429decdaabf286..79580339a0f9620b1b702dc2fa271ff5e9e0add9 100644
--- a/python/paddle/fluid/tests/unittests/test_gather_tree_op.py
+++ b/python/paddle/fluid/tests/unittests/test_gather_tree_op.py
@@ -18,7 +18,6 @@ import numpy as np
 from op_test import OpTest
 
 import paddle
-import paddle.fluid as fluid
 from paddle.fluid.framework import Program, program_guard
 
 
@@ -58,14 +57,11 @@ class TestGatherTreeOp(OpTest):
 class TestGatherTreeOpAPI(unittest.TestCase):
     def test_case(self):
         paddle.enable_static()
-        ids = fluid.layers.data(
-            name='ids', shape=[5, 2, 2], dtype='int64', append_batch_size=False
-        )
-        parents = fluid.layers.data(
+        ids = paddle.static.data(name='ids', shape=[5, 2, 2], dtype='int64')
+        parents = paddle.static.data(
             name='parents',
             shape=[5, 2, 2],
             dtype='int64',
-            append_batch_size=False,
         )
         final_sequences = paddle.nn.functional.gather_tree(ids, parents)
         paddle.disable_static()
@@ -84,17 +80,9 @@ class TestGatherTreeOpError(unittest.TestCase):
     def test_errors(self):
         paddle.enable_static()
         with program_guard(Program(), Program()):
-            ids = fluid.layers.data(
-                name='ids',
-                shape=[5, 2, 2],
-                dtype='int64',
-                append_batch_size=False,
-            )
-            parents = fluid.layers.data(
-                name='parents',
-                shape=[5, 2, 2],
-                dtype='int64',
-                append_batch_size=False,
+            ids = paddle.static.data(name='ids', shape=[5, 2, 2], dtype='int64')
+            parents = paddle.static.data(
+                name='parents', shape=[5, 2, 2], dtype='int64'
             )
 
             def test_Variable_ids():
@@ -113,11 +101,8 @@ class TestGatherTreeOpError(unittest.TestCase):
 
             def test_type_ids():
                 # dtype must be int32 or int64
-                bad_ids = fluid.layers.data(
-                    name='bad_ids',
-                    shape=[5, 2, 2],
-                    dtype='float32',
-                    append_batch_size=False,
+                bad_ids = paddle.static.data(
+                    name='bad_ids', shape=[5, 2, 2], dtype='float32'
                 )
                 paddle.nn.functional.gather_tree(bad_ids, parents)
 
@@ -125,33 +110,24 @@ class TestGatherTreeOpError(unittest.TestCase):
 
             def test_type_parents():
                 # dtype must be int32 or int64
-                bad_parents = fluid.layers.data(
-                    name='bad_parents',
-                    shape=[5, 2, 2],
-                    dtype='float32',
-                    append_batch_size=False,
+                bad_parents = paddle.static.data(
+                    name='bad_parents', shape=[5, 2, 2], dtype='float32'
                 )
                 paddle.nn.functional.gather_tree(ids, bad_parents)
 
             self.assertRaises(TypeError, test_type_parents)
 
             def test_ids_ndim():
-                bad_ids = fluid.layers.data(
-                    name='bad_test_ids',
-                    shape=[5, 2],
-                    dtype='int64',
-                    append_batch_size=False,
+                bad_ids = paddle.static.data(
+                    name='bad_test_ids', shape=[5, 2], dtype='int64'
                 )
                 paddle.nn.functional.gather_tree(bad_ids, parents)
 
             self.assertRaises(ValueError, test_ids_ndim)
 
             def test_parents_ndim():
-                bad_parents = fluid.layers.data(
-                    name='bad_test_parents',
-                    shape=[5, 2],
-                    dtype='int64',
-                    append_batch_size=False,
+                bad_parents = paddle.static.data(
+                    name='bad_test_parents', shape=[5, 2], dtype='int64'
                 )
                 paddle.nn.functional.gather_tree(ids, bad_parents)
 
diff --git a/python/paddle/fluid/tests/unittests/test_generator_dataloader.py b/python/paddle/fluid/tests/unittests/test_generator_dataloader.py
index 6d7b9914257620e20e001e277190a1f2f9ed97e3..fcbc91edee31eec09cb8cd57b94a19a7d39d7030 100644
--- a/python/paddle/fluid/tests/unittests/test_generator_dataloader.py
+++ b/python/paddle/fluid/tests/unittests/test_generator_dataloader.py
@@ -43,10 +43,12 @@ def simple_fc_net(places, use_legacy_py_reader, use_double_buffer):
 
     with fluid.unique_name.guard():
         with fluid.program_guard(main_prog, startup_prog):
-            image = fluid.layers.data(
-                name='image', shape=[784], dtype='float32'
+            image = paddle.static.data(
+                name='image', shape=[-1, 784], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             py_reader = fluid.io.DataLoader.from_generator(
                 feed_list=[image, label],
                 capacity=4,
diff --git a/python/paddle/fluid/tests/unittests/test_group_norm_op.py b/python/paddle/fluid/tests/unittests/test_group_norm_op.py
index 52a233658a64e7e43eae5e1ab12160259dd5984e..01cd8108a3c8e001dde5ee6b9e5bd25c6906c313 100644
--- a/python/paddle/fluid/tests/unittests/test_group_norm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_group_norm_op.py
@@ -52,8 +52,8 @@ class TestGroupNormOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_x_type)
 
             def test_x_dtype():
-                x2 = fluid.layers.data(
-                    name='x2', shape=[2, 100, 3, 5], dtype='int32'
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 2, 100, 3, 5], dtype='int32'
                 )
                 groups = 2
                 paddle.static.nn.group_norm(x2, groups)
diff --git a/python/paddle/fluid/tests/unittests/test_hsigmoid_op.py b/python/paddle/fluid/tests/unittests/test_hsigmoid_op.py
index 6c5bc338a565b87a47bac3184ca97f4fed4483d6..75e5d1ee2ee158528bcc120b347614c0a3d3dc59 100644
--- a/python/paddle/fluid/tests/unittests/test_hsigmoid_op.py
+++ b/python/paddle/fluid/tests/unittests/test_hsigmoid_op.py
@@ -286,14 +286,14 @@ class TestHSigmoidOpSparse(OpTest):
 
 class TestHSigmoidOpWithSparseGrad(unittest.TestCase):
     def hs_net_conf(self, is_sparse):
-        input_word = fluid.layers.data(name="x", shape=[1], dtype='int64')
-        path_table = fluid.layers.data(
-            name='path_table', shape=[3], dtype='int64'
+        input_word = paddle.static.data(name="x", shape=[-1, 1], dtype='int64')
+        path_table = paddle.static.data(
+            name='path_table', shape=[-1, 3], dtype='int64'
         )
-        path_code = fluid.layers.data(
-            name='path_code', shape=[3], dtype='int64'
+        path_code = paddle.static.data(
+            name='path_code', shape=[-1, 3], dtype='int64'
         )
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
         data_list = [input_word, path_table, path_code, label]
 
diff --git a/python/paddle/fluid/tests/unittests/test_identity_loss_op.py b/python/paddle/fluid/tests/unittests/test_identity_loss_op.py
index 1a0ff98b176e55c214792f22ae679ee28b030335..d9b8ee8fad32812adf92194fec493fcddba361cc 100644
--- a/python/paddle/fluid/tests/unittests/test_identity_loss_op.py
+++ b/python/paddle/fluid/tests/unittests/test_identity_loss_op.py
@@ -103,7 +103,7 @@ class TestIdentityLossOpError(unittest.TestCase):
             self.assertRaises(Exception, test_string)
 
             def test_dtype():
-                x2 = fluid.layers.data(name='x2', shape=[1], dtype='int32')
+                x2 = paddle.static.data(name='x2', shape=[-1, 1], dtype='int32')
                 paddle.incubate.identity_loss(x=x2, reduction=1)
 
             self.assertRaises(TypeError, test_dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_image_classification_layer.py b/python/paddle/fluid/tests/unittests/test_image_classification_layer.py
index c485a5182843447d051cf4406304eb91849e3b3e..7ffa8f5e53cff97484c360678fe36e55501d1f0b 100644
--- a/python/paddle/fluid/tests/unittests/test_image_classification_layer.py
+++ b/python/paddle/fluid/tests/unittests/test_image_classification_layer.py
@@ -39,8 +39,8 @@ class TestLayer(unittest.TestCase):
         main_program = Program()
         startup_program = Program()
         with fluid.program_guard(main_program, startup_program):
-            images = fluid.layers.data(
-                name='pixel', shape=[3, 48, 48], dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 3, 48, 48], dtype='float32'
             )
             hidden1 = paddle.static.nn.batch_norm(input=images)
             hidden2 = paddle.static.nn.fc(
@@ -54,8 +54,8 @@ class TestLayer(unittest.TestCase):
         main_program = Program()
         startup_program = Program()
         with fluid.program_guard(main_program, startup_program):
-            images = fluid.layers.data(
-                name='pixel', shape=[3, 48, 48], dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 3, 48, 48], dtype='float32'
             )
             paddle.nn.functional.dropout(x=images, p=0.5)
 
@@ -66,8 +66,8 @@ class TestLayer(unittest.TestCase):
         startup_program = Program()
 
         with fluid.program_guard(main_program, startup_program):
-            images = fluid.layers.data(
-                name='pixel', shape=[3, 48, 48], dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 3, 48, 48], dtype='float32'
             )
             conv1 = conv_block(images, 64, 2, [0.3, 0])
             conv_block(conv1, 256, 3, [0.4, 0.4, 0])
@@ -78,11 +78,11 @@ class TestLayer(unittest.TestCase):
         main_program = Program()
         startup_program = Program()
         with fluid.program_guard(main_program, startup_program):
-            image1 = fluid.layers.data(
-                name='pixel1', shape=[3, 48, 48], dtype='float32'
+            image1 = paddle.static.data(
+                name='pixel1', shape=[-1, 3, 48, 48], dtype='float32'
             )
-            image2 = fluid.layers.data(
-                name='pixel2', shape=[3, 48, 48], dtype='float32'
+            image2 = paddle.static.data(
+                name='pixel2', shape=[-1, 3, 48, 48], dtype='float32'
             )
             paddle.nn.functional.relu(paddle.add(x=image1, y=image2))
         print(main_program)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_deepcf.py b/python/paddle/fluid/tests/unittests/test_imperative_deepcf.py
index 21f95491fa69e9ae8863d1bbed62a4783b3a5639..f34c8d6a2a85832a6c3490ebfe05aa9f63f16189 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_deepcf.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_deepcf.py
@@ -264,9 +264,9 @@ class TestDygraphDeepCF(unittest.TestCase):
 
         scope = fluid.core.Scope()
         with new_program_scope(main=main, startup=startup, scope=scope):
-            users = fluid.layers.data('users', [1], dtype='int32')
-            items = fluid.layers.data('items', [1], dtype='int32')
-            labels = fluid.layers.data('labels', [1], dtype='float32')
+            users = paddle.static.data('users', [-1, 1], dtype='int32')
+            items = paddle.static.data('items', [-1, 1], dtype='int32')
+            labels = paddle.static.data('labels', [-1, 1], dtype='float32')
 
             deepcf = DeepCF(num_users, num_items, matrix)
             prediction = deepcf(users, items)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_framework.py b/python/paddle/fluid/tests/unittests/test_imperative_framework.py
index c0676fcc8524fd579c0647f2399dace335ece061..a14a508c6888e61176dc3260438e0a3fff60a161 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_framework.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_framework.py
@@ -56,9 +56,7 @@ class TestDygraphFramework(unittest.TestCase):
     def test_dygraph_backward(self):
         with new_program_scope():
             mlp = MLP(input_size=2)
-            var_inp = fluid.layers.data(
-                "input", shape=[2, 2], dtype="float32", append_batch_size=False
-            )
+            var_inp = paddle.static.data("input", shape=[2, 2], dtype="float32")
             out = mlp(var_inp)
             try:
                 out.backward()
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_gan.py b/python/paddle/fluid/tests/unittests/test_imperative_gan.py
index 0ab0703b16a09e2bb22238d7a30659ce720c2f1a..34806a8305a7246c457ac034f6a6af065727c6b3 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_gan.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_gan.py
@@ -70,12 +70,8 @@ class TestDygraphGAN(unittest.TestCase):
             discriminator = Discriminator()
             generator = Generator()
 
-            img = fluid.layers.data(
-                name="img", shape=[2, 1], append_batch_size=False
-            )
-            noise = fluid.layers.data(
-                name="noise", shape=[2, 2], append_batch_size=False
-            )
+            img = paddle.static.data(name="img", shape=[2, 1])
+            noise = paddle.static.data(name="noise", shape=[2, 2])
 
             d_real = discriminator(img)
             d_loss_real = paddle.mean(
@@ -106,9 +102,7 @@ class TestDygraphGAN(unittest.TestCase):
             discriminator = Discriminator()
             generator = Generator()
 
-            noise = fluid.layers.data(
-                name="noise", shape=[2, 2], append_batch_size=False
-            )
+            noise = paddle.static.data(name="noise", shape=[2, 2])
 
             d_fake = discriminator(generator(noise))
             g_loss = paddle.mean(
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_gnn.py b/python/paddle/fluid/tests/unittests/test_imperative_gnn.py
index 54dffdf72362182b81015f3e2a9b9b1927bf4d5b..25eafb2daaf566c25800771dfcb6bf32bee5e4a0 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_gnn.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_gnn.py
@@ -71,24 +71,15 @@ class TestDygraphGNN(unittest.TestCase):
 
         scope = fluid.core.Scope()
         with new_program_scope(main=main, startup=startup, scope=scope):
-            features = fluid.layers.data(
-                name='features',
-                shape=[1, 100, 50],
-                dtype='float32',
-                append_batch_size=False,
+            features = paddle.static.data(
+                name='features', shape=[1, 100, 50], dtype='float32'
             )
             # Use selected rows when it's supported.
-            adj = fluid.layers.data(
-                name='adj',
-                shape=[1, 100, 100],
-                dtype='float32',
-                append_batch_size=False,
+            adj = paddle.static.data(
+                name='adj', shape=[1, 100, 100], dtype='float32'
             )
-            labels = fluid.layers.data(
-                name='labels',
-                shape=[100, 1],
-                dtype='int64',
-                append_batch_size=False,
+            labels = paddle.static.data(
+                name='labels', shape=[100, 1], dtype='int64'
             )
 
             model = GCN('test_gcn', 50)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_lod_tensor_to_selected_rows.py b/python/paddle/fluid/tests/unittests/test_imperative_lod_tensor_to_selected_rows.py
index 0a775f75e71dac18cb4af16b6960227e0c8bd59c..0eb037bc6a02e40d8f2734dfd0b051ad996d4204 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_lod_tensor_to_selected_rows.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_lod_tensor_to_selected_rows.py
@@ -163,11 +163,12 @@ class TestDygraphSimpleNet(unittest.TestCase):
 
                     exe = fluid.Executor(place)
                     sgd = SGDOptimizer(learning_rate=1e-3)
-                    x = fluid.layers.data(
+                    x = paddle.static.data(
                         name="x", shape=[-1, num_steps], dtype='int64'
                     )
-                    y = fluid.layers.data(name="y", shape=[-1, 1], dtype=dtype)
-
+                    x.desc.set_need_check_feed(False)
+                    y = paddle.static.data(name="y", shape=[-1, 1], dtype=dtype)
+                    y.desc.set_need_check_feed(False)
                     static_loss = simple_net(x, y)
                     sgd.minimize(static_loss)
                     static_param_updated = dict()
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_mnist.py b/python/paddle/fluid/tests/unittests/test_imperative_mnist.py
index 0e813147ad95504157e1fd8b40d82febd240b6cc..9c187d73b81e48b83cfd60e9a31e853d12d43ab3 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_mnist.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_mnist.py
@@ -195,10 +195,12 @@ class TestImperativeMnist(unittest.TestCase):
                 drop_last=True,
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             cost = mnist(img)
             loss = paddle.nn.functional.cross_entropy(
                 cost, label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_mnist_sorted_gradient.py b/python/paddle/fluid/tests/unittests/test_imperative_mnist_sorted_gradient.py
index 64594acb285b027eaef1161c00abbe7ebccd4a45..3acf62766165ffc471d865f559eddabd63800d75 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_mnist_sorted_gradient.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_mnist_sorted_gradient.py
@@ -98,10 +98,12 @@ class TestImperativeMnistSortGradient(unittest.TestCase):
                 paddle.dataset.mnist.train(), batch_size=128, drop_last=True
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             cost = mnist(img)
             loss = paddle.nn.functional.cross_entropy(
                 cost, label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_ocr_attention_model.py b/python/paddle/fluid/tests/unittests/test_imperative_ocr_attention_model.py
index 6217ef0b1b5bfdaf06f76154878c16349dc6489f..12be3af2d9cf9164a7a865e4e4de79fd33299a22 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_ocr_attention_model.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_ocr_attention_model.py
@@ -537,15 +537,19 @@ class TestDygraphOCRAttention(unittest.TestCase):
 
             optimizer = fluid.optimizer.SGD(learning_rate=0.001)
 
-            images = fluid.layers.data(
-                name='pixel', shape=Config.DATA_SHAPE, dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1] + Config.DATA_SHAPE, dtype='float32'
             )
-            static_label_in = fluid.layers.data(
-                name='label_in', shape=[1], dtype='int64', lod_level=0
+            images.desc.set_need_check_feed(False)
+            static_label_in = paddle.static.data(
+                name='label_in', shape=[-1, 1], dtype='int64', lod_level=0
             )
-            static_label_out = fluid.layers.data(
-                name='label_out', shape=[1], dtype='int64', lod_level=0
+            static_label_in.desc.set_need_check_feed(False)
+            static_label_out = paddle.static.data(
+                name='label_out', shape=[-1, 1], dtype='int64', lod_level=0
             )
+            static_label_out.desc.set_need_check_feed(False)
+
             static_label_out.stop_gradient = True
             static_label_out.trainable = False
 
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_optimizer.py b/python/paddle/fluid/tests/unittests/test_imperative_optimizer.py
index ab1490588088908abdc4938fce9a37577fdd3d26..31f066a1f186518d9f029e6423a428b4bf08f1ff 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_optimizer.py
@@ -173,10 +173,12 @@ class TestImperativeOptimizerBase(unittest.TestCase):
                 paddle.dataset.mnist.train(), batch_size=128, drop_last=True
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             img = paddle.reshape(img, shape=[batch_size, 784])
             cost = mlp(img)
             avg_loss = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_optimizer_v2.py b/python/paddle/fluid/tests/unittests/test_imperative_optimizer_v2.py
index 2246cc25a22285ce87162613f24c0c96fc0d49aa..d038e89e1fb532fda882865f3e042d466c9ef855 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_optimizer_v2.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_optimizer_v2.py
@@ -182,10 +182,12 @@ class TestImperativeOptimizerBase(unittest.TestCase):
                 paddle.dataset.mnist.train(), batch_size=128, drop_last=True
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[1, 28, 28], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 1, 28, 28], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             img = paddle.reshape(img, shape=[batch_size, 784])
             cost = mlp(img)
             avg_loss = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn.py b/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn.py
index b24b24ebc6255527a4b6dd575f86fe044623f64e..6eb0c9d6e6c0345d9cedd8273c5e4b2bbaeacee6 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn.py
@@ -328,16 +328,20 @@ class TestDygraphPtbRnn(unittest.TestCase):
                 else fluid.CUDAPlace(0)
             )
             sgd = SGDOptimizer(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn_sorted_gradient.py b/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn_sorted_gradient.py
index 01b1d18070e1480ae22f493ecad9ac0afa7373ff..a586c98de1e075e134e4c5ce41821fde5fabf463 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn_sorted_gradient.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_ptb_rnn_sorted_gradient.py
@@ -116,16 +116,20 @@ class TestDygraphPtbRnnSortGradient(unittest.TestCase):
                 else fluid.CUDAPlace(0)
             )
             sgd = SGDOptimizer(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps, 1], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_recurrent_usage.py b/python/paddle/fluid/tests/unittests/test_imperative_recurrent_usage.py
index e40cb632065fa58e66a9c40ee95921e4901b5d2b..2d57340b954eeeca3628dc2868715e2f677466c6 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_recurrent_usage.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_recurrent_usage.py
@@ -80,12 +80,8 @@ class TestRecurrentFeed(unittest.TestCase):
         with new_program_scope():
             fluid.default_startup_program().random_seed = seed
             fluid.default_main_program().random_seed = seed
-            in1 = fluid.layers.data(
-                name="inp1", shape=[2, 2], append_batch_size=False
-            )
-            in2 = fluid.layers.data(
-                name="inp2", shape=[2, 2], append_batch_size=False
-            )
+            in1 = paddle.static.data(name="inp1", shape=[2, 2])
+            in2 = paddle.static.data(name="inp2", shape=[2, 2])
             rt1 = RecurrentTest("RecurrentTest")
             static_sum_out, static_out = rt1(in1, in2)
             fluid.backward.append_backward(static_sum_out)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_reinforcement.py b/python/paddle/fluid/tests/unittests/test_imperative_reinforcement.py
index 1c548f5caea48863c31a511c90b88f4806e5954b..88440c0cf22d163e31f63d385ad6ec2bf75a9018 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_reinforcement.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_reinforcement.py
@@ -126,14 +126,14 @@ class TestImperativeMnist(unittest.TestCase):
 
             st_sgd = SGDOptimizer(learning_rate=1e-3)
 
-            st_state = fluid.layers.data(
-                name='st_state', shape=[4], dtype='float32'
+            st_state = paddle.static.data(
+                name='st_state', shape=[-1, 4], dtype='float32'
             )
-            st_reward = fluid.layers.data(
-                name='st_reward', shape=[1], dtype='float32'
+            st_reward = paddle.static.data(
+                name='st_reward', shape=[-1, 1], dtype='float32'
             )
-            st_mask = fluid.layers.data(
-                name='st_mask', shape=[2], dtype='float32'
+            st_mask = paddle.static.data(
+                name='st_mask', shape=[-1, 2], dtype='float32'
             )
 
             st_loss_probs = policy(st_state)
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_resnet.py b/python/paddle/fluid/tests/unittests/test_imperative_resnet.py
index b20ceb091b017f9e55d8bf6aace2086831607d61..bc46ad12d3df0d7c78f1f7085f557d58ca4faa6f 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_resnet.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_resnet.py
@@ -361,10 +361,12 @@ class TestDygraphResnet(unittest.TestCase):
                 batch_size=batch_size,
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[3, 224, 224], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 224, 224], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             out = resnet(img)
             loss = paddle.nn.functional.cross_entropy(
                 input=out, label=label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_resnet_sorted_gradient.py b/python/paddle/fluid/tests/unittests/test_imperative_resnet_sorted_gradient.py
index 798890a4898e8aef61faaa0ba245f525d0d2f471..9d6d2ebabd12ee61721186b8cfef0ea763426285 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_resnet_sorted_gradient.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_resnet_sorted_gradient.py
@@ -170,10 +170,12 @@ class TestDygraphResnetSortGradient(unittest.TestCase):
                 batch_size=batch_size,
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[3, 224, 224], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 224, 224], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             out = resnet(img)
             loss = paddle.nn.functional.cross_entropy(
                 input=out, label=label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_se_resnext.py b/python/paddle/fluid/tests/unittests/test_imperative_se_resnext.py
index b02e1595dbe1ca150fdcedc80c2424a06c60b72c..6c95805602f64a525c938c0e55d39acc6fefadcf 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_se_resnext.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_se_resnext.py
@@ -444,10 +444,12 @@ class TestImperativeResneXt(unittest.TestCase):
                 drop_last=True,
             )
 
-            img = fluid.layers.data(
-                name='pixel', shape=[3, 224, 224], dtype='float32'
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 224, 224], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             out = se_resnext(img)
             softmax_out = paddle.nn.function.softmax(out)
             loss = paddle.nn.functional.cross_entropy(
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_selected_rows_to_lod_tensor.py b/python/paddle/fluid/tests/unittests/test_imperative_selected_rows_to_lod_tensor.py
index cd707bb0ca6420139e500bd21d794e6802cd1bbf..647710fba61f1f4a0eef33840431cfa58e9b5544 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_selected_rows_to_lod_tensor.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_selected_rows_to_lod_tensor.py
@@ -173,11 +173,12 @@ class TestDygraphSimpleNet(unittest.TestCase):
 
                     exe = fluid.Executor(place)
                     sgd = SGDOptimizer(learning_rate=1e-3)
-                    x = fluid.layers.data(
+                    x = paddle.static.data(
                         name="x", shape=[-1, num_steps], dtype='int64'
                     )
-                    y = fluid.layers.data(name="y", shape=[-1, 1], dtype=dtype)
-
+                    x.desc.set_need_check_feed(False)
+                    y = paddle.static.data(name="y", shape=[-1, 1], dtype=dtype)
+                    y.desc.set_need_check_feed(False)
                     static_loss = simple_net(x, y)
                     sgd.minimize(static_loss)
                     static_param_updated = dict()
diff --git a/python/paddle/fluid/tests/unittests/test_imperative_transformer_sorted_gradient.py b/python/paddle/fluid/tests/unittests/test_imperative_transformer_sorted_gradient.py
index 8e213c2d2a2b865fe2384c243a4888fb7fa68785..e171899289aa4be06df5dd9c02d4a1fd4c69f169 100644
--- a/python/paddle/fluid/tests/unittests/test_imperative_transformer_sorted_gradient.py
+++ b/python/paddle/fluid/tests/unittests/test_imperative_transformer_sorted_gradient.py
@@ -221,14 +221,13 @@ def make_all_inputs(input_fields):
     """
     inputs = []
     for input_field in input_fields:
-        input_var = fluid.layers.data(
+        input_var = paddle.static.data(
             name=input_field,
             shape=input_descs[input_field][0],
             dtype=input_descs[input_field][1],
             lod_level=input_descs[input_field][2]
             if len(input_descs[input_field]) == 3
             else 0,
-            append_batch_size=False,
         )
         inputs.append(input_var)
     return inputs
diff --git a/python/paddle/fluid/tests/unittests/test_index_select_op.py b/python/paddle/fluid/tests/unittests/test_index_select_op.py
index 39895f2691ca3f774b60323734d32d3ec4c88491..2234280939cecf37512f6cc927f32101f723222f 100644
--- a/python/paddle/fluid/tests/unittests/test_index_select_op.py
+++ b/python/paddle/fluid/tests/unittests/test_index_select_op.py
@@ -97,10 +97,8 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         # case 1:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
-            index = fluid.layers.data(
-                name='index', shape=[3], dtype='int32', append_batch_size=False
-            )
+            x = paddle.static.data(name='x', shape=[-1, 4])
+            index = paddle.static.data(name='index', shape=[3], dtype='int32')
             z = paddle.index_select(x, index, axis=1)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -115,10 +113,8 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         # case 2:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
-            index = fluid.layers.data(
-                name='index', shape=[3], dtype='int32', append_batch_size=False
-            )
+            x = paddle.static.data(name='x', shape=[-1, 4])
+            index = paddle.static.data(name='index', shape=[3], dtype='int32')
             z = paddle.index_select(x, index)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
diff --git a/python/paddle/fluid/tests/unittests/test_inference_model_io.py b/python/paddle/fluid/tests/unittests/test_inference_model_io.py
index f6b3ba6b69539ca1d36606b4ec2fd38f48b1a938..fd8523bfc15ea5681ec44e0fff95859858e0b23f 100644
--- a/python/paddle/fluid/tests/unittests/test_inference_model_io.py
+++ b/python/paddle/fluid/tests/unittests/test_inference_model_io.py
@@ -24,7 +24,6 @@ import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 import paddle.fluid.executor as executor
-import paddle.fluid.layers as layers
 import paddle.fluid.optimizer as optimizer
 from paddle.distributed.io import load_inference_model_distributed
 from paddle.fluid.compiler import CompiledProgram
@@ -55,8 +54,8 @@ class TestBook(unittest.TestCase):
         program = Program()
 
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
             y_predict = paddle.static.nn.fc(x=x, size=1, activation=None)
 
@@ -163,8 +162,8 @@ class TestSaveInferenceModel(unittest.TestCase):
 
         # fake program without feed/fetch
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
 
@@ -188,8 +187,8 @@ class TestSaveInferenceModel(unittest.TestCase):
 
         # fake program without feed/fetch
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='int32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='int32')
             predict = paddle.static.nn.fc(x, size=2, activation='softmax')
             acc = paddle.static.accuracy(input=predict, label=y)
             auc_var, batch_auc_var, auc_states = paddle.static.auc(
@@ -223,8 +222,8 @@ class TestInstance(unittest.TestCase):
 
         # fake program without feed/fetch
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
 
@@ -261,8 +260,8 @@ class TestSaveInferenceModelNew(unittest.TestCase):
 
         # fake program without feed/fetch
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
 
@@ -440,8 +439,8 @@ class TestSaveInferenceModelNew(unittest.TestCase):
 
         # fake program without feed/fetch
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
 
@@ -489,8 +488,8 @@ class TestSaveInferenceModelNew(unittest.TestCase):
 
         # fake program without feed/fetch
         with program_guard(program, init_program):
-            x = layers.data(name='x', shape=[2], dtype='float32')
-            y = layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
 
diff --git a/python/paddle/fluid/tests/unittests/test_inplace_abn_op.py b/python/paddle/fluid/tests/unittests/test_inplace_abn_op.py
index 299d3218cfdba26ae9094f1441ee7be55bc4d403..a4cc038bd7a52840814f44365f9db5b3b811e4b7 100644
--- a/python/paddle/fluid/tests/unittests/test_inplace_abn_op.py
+++ b/python/paddle/fluid/tests/unittests/test_inplace_abn_op.py
@@ -48,13 +48,13 @@ class TestInplaceANBOpTraining(unittest.TestCase):
         startup.random_seed = seed
         with fluid.unique_name.guard():
             with fluid.program_guard(main, startup):
-                data = fluid.layers.data(
+                data = paddle.static.data(
                     name='input',
                     shape=self.dshape,
                     dtype=self.dtype,
-                    append_batch_size=False,
-                    stop_gradient=False,
                 )
+                data.stop_gradient = False
+                data.desc.set_need_check_feed(False)
 
                 bn = paddle.static.nn.batch_norm(
                     data,
diff --git a/python/paddle/fluid/tests/unittests/test_inplace_softmax_with_cross_entropy.py b/python/paddle/fluid/tests/unittests/test_inplace_softmax_with_cross_entropy.py
index 16477e086daf11a39cbb9e6b4053a65ebb1402ca..b614709ec996673b62a359ca52f69841486ec314 100644
--- a/python/paddle/fluid/tests/unittests/test_inplace_softmax_with_cross_entropy.py
+++ b/python/paddle/fluid/tests/unittests/test_inplace_softmax_with_cross_entropy.py
@@ -37,18 +37,18 @@ class TestSoftmaxWithXe(unittest.TestCase):
         m, n = x.shape
         with fluid.program_guard(fluid.Program(), fluid.Program()):
             with fluid.scope_guard(fluid.Scope()):
-                x_d = fluid.layers.data(
+                x_d = paddle.static.data(
                     name='x',
                     shape=[m, n],
                     dtype=self.dtype,
-                    append_batch_size=False,
                 )
-                y_d = fluid.layers.data(
+                x_d.desc.set_need_check_feed(False)
+                y_d = paddle.static.data(
                     name='y',
                     shape=[m, 1] if not self.soft_label else [m, n],
                     dtype='int64' if not self.soft_label else self.dtype,
-                    append_batch_size=False,
                 )
+                y_d.desc.set_need_check_feed(False)
                 z_d, s_d = paddle.nn.functional.softmax_with_cross_entropy(
                     x_d,
                     y_d,
diff --git a/python/paddle/fluid/tests/unittests/test_instance_norm_op.py b/python/paddle/fluid/tests/unittests/test_instance_norm_op.py
index aac98fec5c43553e035850b6bd57a72fc548b102..48932e5581b24aec2017042df5d4b6d17e6fb2a9 100644
--- a/python/paddle/fluid/tests/unittests/test_instance_norm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_instance_norm_op.py
@@ -243,7 +243,9 @@ class TestInstanceNormOpError(unittest.TestCase):
             self.assertRaises(TypeError, paddle.static.nn.instance_norm, x1)
 
             # the input dtype of instance_norm must be float32 or float64
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
+            )
             self.assertRaises(TypeError, paddle.static.nn.instance_norm, x2)
 
 
@@ -251,9 +253,7 @@ class TestInstanceNormOpErrorCase1(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
             # the first dimension of input for instance_norm must between [2d, 5d]
-            x = fluid.layers.data(
-                name='x', shape=[3], dtype="float32", append_batch_size=False
-            )
+            x = paddle.static.data(name='x', shape=[3], dtype="float32")
             self.assertRaises(ValueError, paddle.static.nn.instance_norm, x)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_ir_inplace_pass.py b/python/paddle/fluid/tests/unittests/test_ir_inplace_pass.py
index 48156b7448218ab67ebfe69191cfa5b02dfa0b84..bc4ef3d386ccb2f0b0c94890b92382867bebe90f 100644
--- a/python/paddle/fluid/tests/unittests/test_ir_inplace_pass.py
+++ b/python/paddle/fluid/tests/unittests/test_ir_inplace_pass.py
@@ -24,8 +24,8 @@ import paddle.fluid.core as core
 
 
 def fc_with_batchnorm(use_feed):
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     hidden = img
     for _ in range(3):
diff --git a/python/paddle/fluid/tests/unittests/test_ir_memory_optimize_pass.py b/python/paddle/fluid/tests/unittests/test_ir_memory_optimize_pass.py
index 9e51118e96c4c9fda4187b1046a9d90f943d4b40..b64090996b3a2fd3de1ccf2bd135fb9b2ac4ea6f 100644
--- a/python/paddle/fluid/tests/unittests/test_ir_memory_optimize_pass.py
+++ b/python/paddle/fluid/tests/unittests/test_ir_memory_optimize_pass.py
@@ -18,13 +18,12 @@ import numpy as np
 from parallel_executor_test_base import DeviceType, TestParallelExecutorBase
 
 import paddle
-import paddle.fluid as fluid
 import paddle.fluid.core as core
 
 
 def _feed_data_helper():
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     return img, label
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_lambv2_op.py b/python/paddle/fluid/tests/unittests/test_lambv2_op.py
index f8f65f5a14a4787bf3522319af80a144739551fc..f7aa3f6fbd92af62499631cc6c0d2fa5611bb37e 100644
--- a/python/paddle/fluid/tests/unittests/test_lambv2_op.py
+++ b/python/paddle/fluid/tests/unittests/test_lambv2_op.py
@@ -123,8 +123,10 @@ class TestLambOpWithCombinedOp(unittest.TestCase):
             with fluid.program_guard(main, startup):
                 main.random_seed = seed
                 startup.random_seed = seed
-                x = fluid.layers.data(name='X', shape=[13], dtype='float32')
-                y = fluid.layers.data(name='Y', shape=[1], dtype='float32')
+                x = paddle.static.data(
+                    name='X', shape=[-1, 13], dtype='float32'
+                )
+                y = paddle.static.data(name='Y', shape=[-1, 1], dtype='float32')
                 prediction = paddle.static.nn.fc(x, size=1, activation=None)
                 loss = paddle.nn.functional.square_error_cost(
                     input=prediction, label=y
diff --git a/python/paddle/fluid/tests/unittests/test_layer_norm_op.py b/python/paddle/fluid/tests/unittests/test_layer_norm_op.py
index 0240b411b047897514e18fc65119464619075ad2..0878f468074500fb23819b36b8667562097356d2 100644
--- a/python/paddle/fluid/tests/unittests/test_layer_norm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_layer_norm_op.py
@@ -333,12 +333,7 @@ class TestLayerNormOp(unittest.TestCase):
 
 class TestLayerNormAPI(unittest.TestCase):
     def test_case(self):
-        x = fluid.layers.data(
-            name='x',
-            shape=[64, 32, 256],
-            dtype='float32',
-            append_batch_size=False,
-        )
+        x = paddle.static.data(name='x', shape=[64, 32, 256], dtype='float32')
         x = paddle.static.nn.layer_norm(
             x,
             scale=True,
@@ -380,7 +375,9 @@ class TestDygraphLayerNormAPIError(unittest.TestCase):
 
             # the input dtype of LayerNorm must be float32 or float64
             # float16 only can be set on GPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 32, 32], dtype="int32")
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 3, 32, 32], dtype="int32"
+            )
             self.assertRaises(TypeError, layer_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_layers.py b/python/paddle/fluid/tests/unittests/test_layers.py
index e33cde31799adc376fe4160e8bfbd027c46b8811..51715e2ae1ce28ecaf6fb161332bbf5144891631 100644
--- a/python/paddle/fluid/tests/unittests/test_layers.py
+++ b/python/paddle/fluid/tests/unittests/test_layers.py
@@ -108,11 +108,10 @@ class TestLayer(LayerTest):
     def test_dropout(self):
         inp = np.ones([3, 32, 32], dtype='float32')
         with self.static_graph():
-            t = layers.data(
+            t = paddle.static.data(
                 name='data',
                 shape=[3, 32, 32],
                 dtype='float32',
-                append_batch_size=False,
             )
             dropout = paddle.nn.Dropout(p=0.35)
             ret = dropout(t)
@@ -135,11 +134,8 @@ class TestLayer(LayerTest):
     def test_linear(self):
         inp = np.ones([3, 32, 32], dtype='float32')
         with self.static_graph():
-            t = layers.data(
-                name='data',
-                shape=[3, 32, 32],
-                dtype='float32',
-                append_batch_size=False,
+            t = paddle.static.data(
+                name='data', shape=[3, 32, 32], dtype='float32'
             )
             linear = paddle.nn.Linear(
                 32, 4, bias_attr=fluid.initializer.ConstantInitializer(value=1)
@@ -228,11 +224,8 @@ class TestLayer(LayerTest):
     def test_Flatten(self):
         inp = np.ones([3, 4, 4, 5], dtype='float32')
         with self.static_graph():
-            t = layers.data(
-                name='data',
-                shape=[3, 4, 4, 5],
-                dtype='float32',
-                append_batch_size=False,
+            t = paddle.static.data(
+                name='data', shape=[3, 4, 4, 5], dtype='float32'
             )
             flatten = paddle.nn.Flatten()
             ret = flatten(t)
@@ -277,7 +270,9 @@ class TestLayer(LayerTest):
     def test_SyncBatchNorm(self):
         if core.is_compiled_with_cuda():
             with self.static_graph():
-                t = layers.data(name='t', shape=[-1, 3, 5, 5], dtype='float32')
+                t = paddle.static.data(
+                    name='t', shape=[-1, 3, 5, 5], dtype='float32'
+                )
                 my_sync_bn = paddle.nn.SyncBatchNorm(3)
                 ret = my_sync_bn(t)
                 static_ret = self.get_static_graph_result(
@@ -294,7 +289,7 @@ class TestLayer(LayerTest):
 
     def test_relu(self):
         with self.static_graph():
-            t = layers.data(name='t', shape=[3, 3], dtype='float32')
+            t = paddle.static.data(name='t', shape=[-1, 3, 3], dtype='float32')
             ret = F.relu(t)
             static_ret = self.get_static_graph_result(
                 feed={'t': np.ones([3, 3], dtype='float32')}, fetch_list=[ret]
@@ -309,8 +304,10 @@ class TestLayer(LayerTest):
 
     def test_matmul(self):
         with self.static_graph():
-            t = layers.data(name='t', shape=[3, 3], dtype='float32')
-            t2 = layers.data(name='t2', shape=[3, 3], dtype='float32')
+            t = paddle.static.data(name='t', shape=[-1, 3, 3], dtype='float32')
+            t2 = paddle.static.data(
+                name='t2', shape=[-1, 3, 3], dtype='float32'
+            )
             ret = paddle.matmul(t, t2)
             static_ret = self.get_static_graph_result(
                 feed={
@@ -337,12 +334,22 @@ class TestLayer(LayerTest):
         n6 = np.ones([3, 3], dtype='float32') * 5
 
         with self.static_graph():
-            t = layers.data(name='t', shape=[3, 3], dtype='float32')
-            t2 = layers.data(name='t2', shape=[3, 3], dtype='float32')
-            t3 = layers.data(name='t3', shape=[3, 3], dtype='float32')
-            t4 = layers.data(name='t4', shape=[3, 3], dtype='float32')
-            t5 = layers.data(name='t5', shape=[3, 3], dtype='float32')
-            t6 = layers.data(name='t6', shape=[3, 3], dtype='float32')
+            t = paddle.static.data(name='t', shape=[-1, 3, 3], dtype='float32')
+            t2 = paddle.static.data(
+                name='t2', shape=[-1, 3, 3], dtype='float32'
+            )
+            t3 = paddle.static.data(
+                name='t3', shape=[-1, 3, 3], dtype='float32'
+            )
+            t4 = paddle.static.data(
+                name='t4', shape=[-1, 3, 3], dtype='float32'
+            )
+            t5 = paddle.static.data(
+                name='t5', shape=[-1, 3, 3], dtype='float32'
+            )
+            t6 = paddle.static.data(
+                name='t6', shape=[-1, 3, 3], dtype='float32'
+            )
 
             ret = paddle.add(t, t2)
             ret = paddle.pow(ret, t3)
@@ -381,7 +388,9 @@ class TestLayer(LayerTest):
     def test_conv2d_transpose(self):
         inp_np = np.arange(0, 24).reshape([2, 3, 2, 2]).astype('float32')
         with self.static_graph():
-            img = layers.data(name='pixel', shape=[3, 2, 2], dtype='float32')
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 2, 2], dtype='float32'
+            )
             out = paddle.static.nn.conv2d_transpose(
                 input=img,
                 num_filters=10,
@@ -393,7 +402,9 @@ class TestLayer(LayerTest):
                 feed={'pixel': inp_np}, fetch_list=[out]
             )[0]
         with self.static_graph():
-            img = layers.data(name='pixel', shape=[3, 2, 2], dtype='float32')
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 2, 2], dtype='float32'
+            )
             conv2d_transpose = paddle.nn.Conv2DTranspose(
                 3,
                 10,
@@ -473,8 +484,8 @@ class TestLayer(LayerTest):
             # the input dtype of Conv2DTranspose must be float16 or float32 or float64
             # float16 only can be set on GPU place
             def test_type():
-                images = layers.data(
-                    name='pixel', shape=[3, 5, 5], dtype='int32'
+                images = paddle.static.data(
+                    name='pixel', shape=[-1, 3, 5, 5], dtype='int32'
                 )
                 conv2d = paddle.nn.Conv2DTranspose(3, 3, [2, 2])
                 conv2d_ret2 = conv2d(images)
@@ -486,12 +497,8 @@ class TestLayer(LayerTest):
         inp_np_y = np.array([[4, 5, 6]]).astype('float32')
 
         with self.static_graph():
-            data_x = layers.data(
-                name='x', shape=[1, 3], dtype="float32", append_batch_size=False
-            )
-            data_y = layers.data(
-                name='y', shape=[1, 3], dtype="float32", append_batch_size=False
-            )
+            data_x = paddle.static.data(name='x', shape=[1, 3], dtype="float32")
+            data_y = paddle.static.data(name='y', shape=[1, 3], dtype="float32")
             out = paddle.static.nn.common.bilinear_tensor_product(
                 data_x,
                 data_y,
@@ -505,12 +512,8 @@ class TestLayer(LayerTest):
             )[0]
 
         with self.static_graph():
-            data_x = layers.data(
-                name='x', shape=[1, 3], dtype="float32", append_batch_size=False
-            )
-            data_y = layers.data(
-                name='y', shape=[1, 3], dtype="float32", append_batch_size=False
-            )
+            data_x = paddle.static.data(name='x', shape=[1, 3], dtype="float32")
+            data_y = paddle.static.data(name='y', shape=[1, 3], dtype="float32")
             btp = paddle.nn.Bilinear(
                 3,
                 3,
@@ -542,11 +545,11 @@ class TestLayer(LayerTest):
             dy_rlt2_value = dy_rlt2.numpy()
 
         with self.static_graph():
-            data_x2 = layers.data(
-                name='x', shape=[1, 3], dtype="float32", append_batch_size=False
+            data_x2 = paddle.static.data(
+                name='x', shape=[1, 3], dtype="float32"
             )
-            data_y2 = layers.data(
-                name='y', shape=[1, 3], dtype="float32", append_batch_size=False
+            data_y2 = paddle.static.data(
+                name='y', shape=[1, 3], dtype="float32"
             )
             out2 = paddle.static.nn.common.bilinear_tensor_product(
                 data_x2, data_y2, 6, act='sigmoid'
@@ -599,7 +602,10 @@ class TestLayer(LayerTest):
         inp_word = np.array([[[1]]]).astype('int64')
         dict_size = 20
         with self.static_graph():
-            data_t = layers.data(name='word', shape=[1], dtype='int64')
+            data_t = paddle.static.data(
+                name='word', shape=[-1, 1], dtype='int64'
+            )
+            data_t.desc.set_need_check_feed(False)
             emb = layers.embedding(
                 input=data_t,
                 size=[dict_size, 32],
@@ -610,7 +616,10 @@ class TestLayer(LayerTest):
                 feed={'word': inp_word}, fetch_list=[emb]
             )[0]
         with self.static_graph():
-            data_t = layers.data(name='word', shape=[1], dtype='int64')
+            data_t = paddle.static.data(
+                name='word', shape=[-1, 1], dtype='int64'
+            )
+            data_t.desc.set_need_check_feed(False)
             emb2 = paddle.nn.Embedding(
                 dict_size, 32, weight_attr='emb.w', sparse=False
             )
@@ -693,8 +702,8 @@ class TestLayer(LayerTest):
 
     def test_conv3d(self):
         with self.static_graph():
-            images = layers.data(
-                name='pixel', shape=[3, 6, 6, 6], dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 3, 6, 6, 6], dtype='float32'
             )
             ret = paddle.static.nn.conv3d(
                 input=images, num_filters=3, filter_size=2
@@ -705,8 +714,8 @@ class TestLayer(LayerTest):
             )[0]
 
         with self.static_graph():
-            images = layers.data(
-                name='pixel', shape=[3, 6, 6, 6], dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 3, 6, 6, 6], dtype='float32'
             )
             conv3d = paddle.nn.Conv3D(
                 in_channels=3, out_channels=3, kernel_size=2
@@ -783,12 +792,8 @@ class TestLayer(LayerTest):
         input = np.random.random(shape).astype('float32')
 
         with self.static_graph():
-            X = fluid.layers.data(
-                name='X',
-                shape=shape,
-                dtype='float32',
-                lod_level=1,
-                append_batch_size=False,
+            X = paddle.static.data(
+                name='X', shape=shape, dtype='float32', lod_level=1
             )
             ret = paddle.static.nn.group_norm(
                 input=X,
@@ -807,12 +812,8 @@ class TestLayer(LayerTest):
             )[0]
 
         with self.static_graph():
-            X = fluid.layers.data(
-                name='X',
-                shape=shape,
-                dtype='float32',
-                lod_level=1,
-                append_batch_size=False,
+            X = paddle.static.data(
+                name='X', shape=shape, dtype='float32', lod_level=1
             )
             groupNorm = paddle.nn.GroupNorm(
                 num_channels=shape[1],
@@ -855,18 +856,14 @@ class TestLayer(LayerTest):
         input = np.random.random(shape).astype('float32')
 
         with self.static_graph():
-            X = fluid.layers.data(
-                name='X', shape=shape, dtype='float32', append_batch_size=False
-            )
+            X = paddle.static.data(name='X', shape=shape, dtype='float32')
             ret = paddle.static.nn.instance_norm(input=X)
             static_ret = self.get_static_graph_result(
                 feed={'X': input}, fetch_list=[ret]
             )[0]
 
         with self.static_graph():
-            X = fluid.layers.data(
-                name='X', shape=shape, dtype='float32', append_batch_size=False
-            )
+            X = paddle.static.data(name='X', shape=shape, dtype='float32')
             instanceNorm = paddle.nn.InstanceNorm2D(num_features=shape[1])
             ret = instanceNorm(X)
             static_ret2 = self.get_static_graph_result(
@@ -914,12 +911,8 @@ class TestLayer(LayerTest):
         input = np.random.random(shape).astype('float32')
 
         with self.static_graph():
-            Weight = fluid.layers.data(
-                name='Weight',
-                shape=shape,
-                dtype='float32',
-                lod_level=1,
-                append_batch_size=False,
+            Weight = paddle.static.data(
+                name='Weight', shape=shape, dtype='float32', lod_level=1
             )
             ret = paddle.static.nn.spectral_norm(
                 weight=Weight, dim=1, power_iters=2
@@ -935,12 +928,8 @@ class TestLayer(LayerTest):
             )[0]
 
         with self.static_graph():
-            Weight = fluid.layers.data(
-                name='Weight',
-                shape=shape,
-                dtype='float32',
-                lod_level=1,
-                append_batch_size=False,
+            Weight = paddle.static.data(
+                name='Weight', shape=shape, dtype='float32', lod_level=1
             )
             spectralNorm = paddle.nn.SpectralNorm(shape, dim=1, power_iters=2)
             ret = spectralNorm(Weight)
@@ -968,7 +957,9 @@ class TestLayer(LayerTest):
         )
 
         with self.static_graph():
-            img = layers.data(name='pixel', shape=[3, 2, 2, 2], dtype='float32')
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 2, 2, 2], dtype='float32'
+            )
             out = paddle.static.nn.conv3d_transpose(
                 input=img, num_filters=12, filter_size=12, use_cudnn=True
             )
@@ -976,7 +967,9 @@ class TestLayer(LayerTest):
                 feed={'pixel': input_array}, fetch_list=[out]
             )[0]
         with self.static_graph():
-            img = layers.data(name='pixel', shape=[3, 2, 2, 2], dtype='float32')
+            img = paddle.static.data(
+                name='pixel', shape=[-1, 3, 2, 2, 2], dtype='float32'
+            )
             conv3d_transpose = paddle.nn.Conv3DTranspose(
                 in_channels=3, out_channels=12, kernel_size=12
             )
@@ -1081,8 +1074,8 @@ class TestLayer(LayerTest):
         value_b = np.arange(3)
         # less than
         with self.static_graph():
-            a = layers.data(name='a', shape=[1], dtype='int64')
-            b = layers.data(name='b', shape=[1], dtype='int64')
+            a = paddle.static.data(name='a', shape=[-1, 1], dtype='int64')
+            b = paddle.static.data(name='b', shape=[-1, 1], dtype='int64')
             cond = paddle.less_than(x=a, y=b)
             static_ret = self.get_static_graph_result(
                 feed={"a": value_a, "b": value_b}, fetch_list=[cond]
@@ -1097,8 +1090,8 @@ class TestLayer(LayerTest):
 
         # less equal
         with self.static_graph():
-            a1 = layers.data(name='a1', shape=[1], dtype='int64')
-            b1 = layers.data(name='b1', shape=[1], dtype='int64')
+            a1 = paddle.static.data(name='a1', shape=[-1, 1], dtype='int64')
+            b1 = paddle.static.data(name='b1', shape=[-1, 1], dtype='int64')
             cond1 = paddle.less_equal(x=a1, y=b1)
             static_ret1 = self.get_static_graph_result(
                 feed={"a1": value_a, "b1": value_b}, fetch_list=[cond1]
@@ -1113,8 +1106,8 @@ class TestLayer(LayerTest):
 
         # greater than
         with self.static_graph():
-            a2 = layers.data(name='a2', shape=[1], dtype='int64')
-            b2 = layers.data(name='b2', shape=[1], dtype='int64')
+            a2 = paddle.static.data(name='a2', shape=[-1, 1], dtype='int64')
+            b2 = paddle.static.data(name='b2', shape=[-1, 1], dtype='int64')
             cond2 = paddle.greater_than(x=a2, y=b2)
             static_ret2 = self.get_static_graph_result(
                 feed={"a2": value_a, "b2": value_b}, fetch_list=[cond2]
@@ -1129,8 +1122,8 @@ class TestLayer(LayerTest):
 
         # greater equal
         with self.static_graph():
-            a3 = layers.data(name='a3', shape=[1], dtype='int64')
-            b3 = layers.data(name='b3', shape=[1], dtype='int64')
+            a3 = paddle.static.data(name='a3', shape=[-1, 1], dtype='int64')
+            b3 = paddle.static.data(name='b3', shape=[-1, 1], dtype='int64')
             cond3 = paddle.greater_equal(x=a3, y=b3)
             static_ret3 = self.get_static_graph_result(
                 feed={"a3": value_a, "b3": value_b}, fetch_list=[cond3]
@@ -1145,8 +1138,8 @@ class TestLayer(LayerTest):
 
         # equal
         with self.static_graph():
-            a4 = layers.data(name='a4', shape=[1], dtype='int64')
-            b4 = layers.data(name='b4', shape=[1], dtype='int64')
+            a4 = paddle.static.data(name='a4', shape=[-1, 1], dtype='int64')
+            b4 = paddle.static.data(name='b4', shape=[-1, 1], dtype='int64')
             cond4 = paddle.equal(x=a4, y=b4)
             static_ret4 = self.get_static_graph_result(
                 feed={"a4": value_a, "b4": value_b}, fetch_list=[cond4]
@@ -1161,8 +1154,8 @@ class TestLayer(LayerTest):
 
         # not equal
         with self.static_graph():
-            a5 = layers.data(name='a5', shape=[1], dtype='int64')
-            b5 = layers.data(name='b5', shape=[1], dtype='int64')
+            a5 = paddle.static.data(name='a5', shape=[-1, 1], dtype='int64')
+            b5 = paddle.static.data(name='b5', shape=[-1, 1], dtype='int64')
             cond5 = paddle.equal(x=a5, y=b5)
             static_ret5 = self.get_static_graph_result(
                 feed={"a5": value_a, "b5": value_b}, fetch_list=[cond5]
@@ -1349,22 +1342,20 @@ class TestLayer(LayerTest):
 
     def test_crop_tensor(self):
         with self.static_graph():
-            x = fluid.layers.data(name="x1", shape=[6, 5, 8])
-
-            dim1 = fluid.layers.data(
-                name="dim1", shape=[1], append_batch_size=False
-            )
-            dim2 = fluid.layers.data(
-                name="dim2", shape=[1], append_batch_size=False
+            x = paddle.static.data(
+                name="x1", shape=[-1, 6, 5, 8], dtype="float32"
             )
+
+            dim1 = paddle.static.data(name="dim1", shape=[1], dtype="float32")
+            dim2 = paddle.static.data(name="dim2", shape=[1], dtype="float32")
             crop_shape1 = (1, 2, 4, 4)
-            crop_shape2 = fluid.layers.data(
-                name="crop_shape", shape=[4], append_batch_size=False
+            crop_shape2 = paddle.static.data(
+                name="crop_shape", shape=[4], dtype="float32"
             )
             crop_shape3 = [-1, dim1, dim2, 4]
             crop_offsets1 = [0, 0, 1, 0]
-            crop_offsets2 = fluid.layers.data(
-                name="crop_offset", shape=[4], append_batch_size=False
+            crop_offsets2 = paddle.static.data(
+                name="crop_offset", shape=[4], dtype="float32"
             )
             crop_offsets3 = [0, dim1, dim2, 0]
 
@@ -1378,7 +1369,9 @@ class TestLayer(LayerTest):
 
     def test_shard_index(self):
         with self.static_graph():
-            x = fluid.layers.data(name="label", shape=[4, 1], dtype='int64')
+            x = paddle.static.data(
+                name="label", shape=[-1, 4, 1], dtype='int64'
+            )
             shard_label = paddle.shard_index(
                 input=x, index_num=20, nshards=2, shard_id=0
             )
@@ -1517,12 +1510,15 @@ class TestBook(LayerTest):
                 self._feed_dict[name] = self._get_np_data(
                     shape, dtype, append_batch_size
                 )
-            return layers.data(
+            if append_batch_size:
+                shape = [-1] + shape
+            data = paddle.static.data(
                 name=name,
                 shape=shape,
                 dtype=dtype,
-                append_batch_size=append_batch_size,
             )
+            data.desc.set_need_check_feed(False)
+            return data
 
     def make_fit_a_line(self):
         with program_guard(
@@ -1803,8 +1799,8 @@ class TestBook(LayerTest):
             updates = self._get_data(
                 name='updates',
                 shape=[2, 3],
-                append_batch_size=False,
                 dtype='float32',
+                append_batch_size=False,
             )
             out = paddle.scatter(x, index=idx, updates=updates)
             return out
@@ -2065,11 +2061,17 @@ class TestBook(LayerTest):
 
     def test_affine_grid(self):
         with self.static_graph():
-            data = layers.data(name='data', shape=[2, 3, 3], dtype="float32")
+            data = paddle.static.data(
+                name='data', shape=[-1, 2, 3, 3], dtype="float32"
+            )
             out = paddle.argsort(x=data, axis=1)
 
-            theta = layers.data(name="theta", shape=[2, 3], dtype="float32")
-            out_shape = layers.data(name="out_shape", shape=[-1], dtype="int32")
+            theta = paddle.static.data(
+                name="theta", shape=[-1, 2, 3], dtype="float32"
+            )
+            out_shape = paddle.static.data(
+                name="out_shape", shape=[-1], dtype="int32"
+            )
             data_0 = paddle.nn.functional.affine_grid(theta, out_shape)
             data_1 = paddle.nn.functional.affine_grid(theta, [5, 3, 28, 28])
 
@@ -2082,7 +2084,9 @@ class TestBook(LayerTest):
         ends = [3, 3, 4]
         strides = [1, 1, 1]
         with self.static_graph():
-            x = layers.data(name="x", shape=[245, 30, 30], dtype="float32")
+            x = paddle.static.data(
+                name="x", shape=[-1, 245, 30, 30], dtype="float32"
+            )
             out = paddle.strided_slice(
                 x, axes=axes, starts=starts, ends=ends, strides=strides
             )
@@ -2101,31 +2105,38 @@ class TestBook(LayerTest):
     def test_sequence_expand(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(name='x', shape=[10], dtype='float32')
-            y = layers.data(
-                name='y', shape=[10, 20], dtype='float32', lod_level=2
+            x = paddle.static.data(name='x', shape=[-1, 10], dtype='float32')
+            y = paddle.static.data(
+                name='y', shape=[-1, 10, 20], dtype='float32', lod_level=2
             )
             return layers.sequence_expand(x=x, y=y, ref_level=1)
 
     def test_sequence_reshape(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(name='x', shape=[8], dtype='float32', lod_level=1)
+            x = paddle.static.data(
+                name='x', shape=[-1, 8], dtype='float32', lod_level=1
+            )
             out = layers.sequence_reshape(input=x, new_dim=16)
             return out
 
     def test_sequence_unpad(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(name='x', shape=[10, 5], dtype='float32')
-            length = layers.data(name='length', shape=[], dtype='int64')
+            x = paddle.static.data(name='x', shape=[-1, 10, 5], dtype='float32')
+            length = paddle.static.data(
+                name='length', shape=[-1], dtype='int64'
+            )
             return layers.sequence_unpad(x=x, length=length)
 
     def test_sequence_softmax(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            seq_data = layers.data(
-                name='seq_data', shape=[10, 10], dtype='float32', lod_level=1
+            seq_data = paddle.static.data(
+                name='seq_data',
+                shape=[-1, 10, 10],
+                dtype='float32',
+                lod_level=1,
             )
             seq = paddle.static.nn.fc(x=seq_data, size=20)
             return layers.sequence_softmax(seq)
@@ -2133,27 +2144,23 @@ class TestBook(LayerTest):
     def test_sequence_unsqueeze(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(name='x', shape=[8, 2], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 8, 2], dtype='float32')
             out = paddle.unsqueeze(x, axis=[1])
             return out
 
     def test_sequence_scatter(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(
-                name='x', shape=[3, 6], append_batch_size=False, dtype='float32'
-            )
-            idx = layers.data(
+            x = paddle.static.data(name='x', shape=[3, 6], dtype='float32')
+            idx = paddle.static.data(
                 name='idx',
                 shape=[12, 1],
-                append_batch_size=False,
                 dtype='int32',
                 lod_level=1,
             )
-            updates = layers.data(
+            updates = paddle.static.data(
                 name='updates',
                 shape=[12, 1],
-                append_batch_size=False,
                 dtype='float32',
                 lod_level=1,
             )
@@ -2165,8 +2172,8 @@ class TestBook(LayerTest):
         with self.static_graph():
             import numpy as np
 
-            seqs = layers.data(
-                name='x', shape=[10, 5], dtype='float32', lod_level=1
+            seqs = paddle.static.data(
+                name='x', shape=[-1, 10, 5], dtype='float32', lod_level=1
             )
             offset = layers.assign(input=np.array([[0, 1]]).astype('int32'))
             length = layers.assign(input=np.array([[2, 1]]).astype('int32'))
@@ -2178,8 +2185,8 @@ class TestBook(LayerTest):
     def test_shuffle_batch(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(
-                name='X', shape=[4, 50], dtype='float32', lod_level=0
+            x = paddle.static.data(
+                name='X', shape=[-1, 4, 50], dtype='float32', lod_level=0
             )
             out1 = fluid.contrib.layers.shuffle_batch(x)
             default_main_program().random_seed = 1000
@@ -2240,21 +2247,25 @@ class TestBook(LayerTest):
     def test_sequence_enumerate(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(name="input", shape=[1], dtype='int32', lod_level=1)
+            x = paddle.static.data(
+                name="input", shape=[-1, 1], dtype='int32', lod_level=1
+            )
             out = layers.sequence_enumerate(input=x, win_size=2, pad_value=0)
 
     def test_row_conv(self):
         # TODO(minqiyang): dygraph do not support lod now
         with self.static_graph():
-            x = layers.data(name='x', shape=[16], dtype='float32', lod_level=1)
+            x = paddle.static.data(
+                name='x', shape=[-1, 16], dtype='float32', lod_level=1
+            )
             out = paddle.static.nn.row_conv(input=x, future_context_size=2)
             return out
 
     def test_simple_conv2d(self):
         # TODO(minqiyang): dygraph do not support layers with param now
         with self.static_graph():
-            images = layers.data(
-                name='pixel', shape=[3, 48, 48], dtype='float32'
+            images = paddle.static.data(
+                name='pixel', shape=[-1, 3, 48, 48], dtype='float32'
             )
             return paddle.static.nn.conv2d(
                 input=images, num_filters=3, filter_size=[4, 4]
@@ -2263,16 +2274,17 @@ class TestBook(LayerTest):
     def test_squeeze(self):
         # TODO(minqiyang): dygraph do not support layers with param now
         with self.static_graph():
-            x = layers.data(name='x', shape=[1, 1, 4], dtype='float32')
+            x = paddle.static.data(
+                name='x', shape=[-1, 1, 1, 4], dtype='float32'
+            )
             out = paddle.squeeze(x, axis=[2])
             return out
 
     def test_flatten(self):
         # TODO(minqiyang): dygraph do not support op without kernel now
         with self.static_graph():
-            x = layers.data(
+            x = paddle.static.data(
                 name='x',
-                append_batch_size=False,
                 shape=[4, 4, 3],
                 dtype="float32",
             )
@@ -2288,7 +2300,9 @@ class TestBook(LayerTest):
 
     def test_unfold(self):
         with self.static_graph():
-            x = layers.data(name='x', shape=[3, 20, 20], dtype='float32')
+            x = paddle.static.data(
+                name='x', shape=[-1, 3, 20, 20], dtype='float32'
+            )
             out = paddle.nn.functional.unfold(x, [3, 3], 1, 1, 1)
             return out
 
@@ -2308,18 +2322,13 @@ class TestBook(LayerTest):
         with program_guard(
             fluid.default_main_program(), fluid.default_startup_program()
         ):
-            input = layers.data(
+            input = paddle.static.data(
                 name='input_data',
                 shape=[3, 3],
-                append_batch_size=False,
                 dtype='float32',
             )
-            x = layers.data(
-                name='x', shape=[3, 2], append_batch_size=False, dtype='float32'
-            )
-            y = layers.data(
-                name='y', shape=[2, 3], append_batch_size=False, dtype='float32'
-            )
+            x = paddle.static.data(name='x', shape=[3, 2], dtype='float32')
+            y = paddle.static.data(name='y', shape=[2, 3], dtype='float32')
 
             out = paddle.addmm(input=input, x=x, y=y)
             return out
diff --git a/python/paddle/fluid/tests/unittests/test_listen_and_serv_op.py b/python/paddle/fluid/tests/unittests/test_listen_and_serv_op.py
index 1ed1a713e30aa3cf3f3fc7a479ea1ec1583a2070..0c1327c5af3a110a6e375c06c2758d1d42ac742d 100644
--- a/python/paddle/fluid/tests/unittests/test_listen_and_serv_op.py
+++ b/python/paddle/fluid/tests/unittests/test_listen_and_serv_op.py
@@ -31,9 +31,9 @@ paddle.enable_static()
 
 def run_pserver(use_cuda, sync_mode, ip, port, trainers, trainer_id):
     remove_ps_flag(os.getpid())
-    x = fluid.layers.data(name='x', shape=[1], dtype='float32')
+    x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
     y_predict = paddle.static.nn.fc(x, size=1, activation=None)
-    y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+    y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
     # loss function
     cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
@@ -68,9 +68,9 @@ def run_pserver_with_empty_block(
     use_cuda, sync_mode, ip, port, trainers, trainer_id
 ):
     remove_ps_flag(os.getpid())
-    x = fluid.layers.data(name='x', shape=[1], dtype='float32')
+    x = paddle.static.data(name='x', shape=[-1, 1], dtype='float32')
     y_predict = paddle.static.nn.fc(x, size=1, bias_attr=False)
-    y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+    y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
     # loss function
     cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
diff --git a/python/paddle/fluid/tests/unittests/test_load_vars_shape_check.py b/python/paddle/fluid/tests/unittests/test_load_vars_shape_check.py
index 3b2465a344ee1f96e5d02a6ddfb3a7e38a620aa5..853b364f358ab1be9504cfd31a8483389ef0bbcb 100644
--- a/python/paddle/fluid/tests/unittests/test_load_vars_shape_check.py
+++ b/python/paddle/fluid/tests/unittests/test_load_vars_shape_check.py
@@ -30,7 +30,9 @@ class TestLoadVarsShapeCheck(unittest.TestCase):
         startup_program_1 = fluid.Program()
 
         with fluid.program_guard(program_1, startup_program_1):
-            input = fluid.layers.data(name="x", shape=[-1, 10], dtype='float32')
+            input = paddle.static.data(
+                name="x", shape=[-1, 10], dtype='float32'
+            )
             out = paddle.static.nn.fc(input, 20)
         place = fluid.CPUPlace()
         exe = Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/test_lookup_table_bf16_op.py b/python/paddle/fluid/tests/unittests/test_lookup_table_bf16_op.py
index ec9eee17c0141600651bd4074667042b64a856f9..cc11e96f5a9150502d20d2b2d25872ed8603ba11 100644
--- a/python/paddle/fluid/tests/unittests/test_lookup_table_bf16_op.py
+++ b/python/paddle/fluid/tests/unittests/test_lookup_table_bf16_op.py
@@ -16,6 +16,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 from paddle import enable_static
@@ -233,7 +234,9 @@ class TestEmbeddingLayerBF16ConstantInitializer(unittest.TestCase):
         self.set_initializer()
 
         with fluid.program_guard(self.prog, self.startup_prog):
-            x = fluid.layers.data(name='x', shape=self.ids_shape, dtype='int64')
+            x = paddle.static.data(
+                name='x', shape=self.ids_shape, dtype='int64'
+            )
             self.emb = fluid.layers.embedding(
                 input=x,
                 size=self.w_shape,
diff --git a/python/paddle/fluid/tests/unittests/test_lookup_table_v2_bf16_op.py b/python/paddle/fluid/tests/unittests/test_lookup_table_v2_bf16_op.py
index c022f564809a981ad8dd8268b349f0a068db4606..8cbc6242b3af985fe497f0d3f8b283d24db49175 100644
--- a/python/paddle/fluid/tests/unittests/test_lookup_table_v2_bf16_op.py
+++ b/python/paddle/fluid/tests/unittests/test_lookup_table_v2_bf16_op.py
@@ -102,7 +102,9 @@ class TestEmbeddingLayerBF16ConstantInitializer(unittest.TestCase):
         self.set_initializer()
 
         with fluid.program_guard(self.prog, self.startup_prog):
-            x = fluid.layers.data(name='x', shape=self.ids_shape, dtype='int64')
+            x = paddle.static.data(
+                name='x', shape=[-1] + self.ids_shape, dtype='int64'
+            )
             self.emb = fluid.input.embedding(
                 input=x,
                 size=self.w_shape,
diff --git a/python/paddle/fluid/tests/unittests/test_lookup_table_v2_op.py b/python/paddle/fluid/tests/unittests/test_lookup_table_v2_op.py
index cea6858e0d4d31aaf3b6f21597bbc8635af63937..eb9c4c60893e06507e9733567b60f9a8aff92e05 100644
--- a/python/paddle/fluid/tests/unittests/test_lookup_table_v2_op.py
+++ b/python/paddle/fluid/tests/unittests/test_lookup_table_v2_op.py
@@ -200,8 +200,8 @@ class TestLookupTableIsSparse(unittest.TestCase):
         self.init_data()
         main_program = fluid.Program()
         with fluid.program_guard(main_program, fluid.Program()):
-            x = fluid.layers.data(name='x', shape=[5], dtype='int64')
-            y_ = fluid.layers.data(name='y_', shape=[5], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 5], dtype='int64')
+            y_ = paddle.static.data(name='y_', shape=[-1, 5], dtype='float32')
             emb = fluid.input.embedding(
                 input=x,
                 size=[10, 16],
@@ -246,7 +246,7 @@ class TestLookupTableIsSparse(unittest.TestCase):
 
 class TestLookupTableApi(unittest.TestCase):
     def test_api(self):
-        x = fluid.layers.data(name='x', shape=[20], dtype='int64')
+        x = paddle.static.data(name='x', shape=[-1, 20], dtype='int64')
         emb = fluid.embedding(input=x, size=[128, 64])
 
         place = fluid.CPUPlace()
diff --git a/python/paddle/fluid/tests/unittests/test_math_op_patch.py b/python/paddle/fluid/tests/unittests/test_math_op_patch.py
index 41cd39088f379a000afdf20a556545b93950d224..4bf9933ca39cdf4ecb78b42fcea93dbe81679f88 100644
--- a/python/paddle/fluid/tests/unittests/test_math_op_patch.py
+++ b/python/paddle/fluid/tests/unittests/test_math_op_patch.py
@@ -28,7 +28,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_add_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = a + 10
         ab = fluid.layers.concat(input=[a, b], axis=1)
         c = ab + 10
@@ -48,7 +48,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_radd_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = 10 + a
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -60,7 +60,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_sub_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = a - 10
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -72,7 +72,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_rsub_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = 10 - a
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -84,7 +84,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_mul_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = a * 10
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -96,7 +96,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_rmul_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = 10 * a
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -108,7 +108,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_div_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = a / 10
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -120,7 +120,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_rdiv_scalar(self):
-        a = fluid.layers.data(name="a", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
         b = 10 / a
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -133,8 +133,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_div_two_tensor(self):
-        a = fluid.layers.data(name="a", shape=[1])
-        b = fluid.layers.data(name="b", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
+        b = paddle.static.data(name="b", shape=[-1, 1])
         c = a / b
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -149,8 +149,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_mul_two_tensor(self):
-        a = fluid.layers.data(name="a", shape=[1])
-        b = fluid.layers.data(name="b", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
+        b = paddle.static.data(name="b", shape=[-1, 1])
         c = a * b
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -165,8 +165,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_add_two_tensor(self):
-        a = fluid.layers.data(name="a", shape=[1])
-        b = fluid.layers.data(name="b", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
+        b = paddle.static.data(name="b", shape=[-1, 1])
         c = a + b
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -181,8 +181,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_sub_two_tensor(self):
-        a = fluid.layers.data(name="a", shape=[1])
-        b = fluid.layers.data(name="b", shape=[1])
+        a = paddle.static.data(name="a", shape=[-1, 1])
+        b = paddle.static.data(name="b", shape=[-1, 1])
         c = a - b
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -197,7 +197,7 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_integer_div(self):
-        a = fluid.layers.data(name="a", shape=[1], dtype='int64')
+        a = paddle.static.data(name="a", shape=[-1, 1], dtype='int64')
         b = a / 7
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -211,8 +211,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_equal(self):
-        a = fluid.layers.data(name="a", shape=[1], dtype='float32')
-        b = fluid.layers.data(name="b", shape=[1], dtype='float32')
+        a = paddle.static.data(name="a", shape=[-1, 1], dtype='float32')
+        b = paddle.static.data(name="b", shape=[-1, 1], dtype='float32')
         c = a == b
 
         place = fluid.CPUPlace()
@@ -231,9 +231,10 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_equal_and_cond(self):
-        a = fluid.layers.data(name="a", shape=[1], dtype='float32')
-        b = fluid.layers.data(name="b", shape=[1], dtype='float32')
-
+        a = paddle.static.data(name="a", shape=[-1, 1], dtype='float32')
+        a.desc.set_need_check_feed(False)
+        b = paddle.static.data(name="b", shape=[-1, 1], dtype='float32')
+        b.desc.set_need_check_feed(False)
         one = paddle.ones(shape=[1], dtype='int32')
         zero = fluid.layers.zeros(shape=[1], dtype='int32')
         cond = one == zero
@@ -241,8 +242,8 @@ class TestMathOpPatches(unittest.TestCase):
 
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
-        a_np = np.array([3, 4, 10, 14, 9, 18]).astype('float')
-        b_np = np.array([3, 4, 11, 15, 8, 18]).astype('float')
+        a_np = np.array([3, 4, 10, 14, 9, 18]).astype('float32')
+        b_np = np.array([3, 4, 11, 15, 8, 18]).astype('float32')
         (c_np,) = exe.run(
             fluid.default_main_program(),
             feed={"a": a_np, "b": b_np},
@@ -253,7 +254,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_neg(self):
-        a = fluid.layers.data(name="a", shape=[10, 1])
+        a = paddle.static.data(name="a", shape=[-1, 10, 1], dtype='float32')
+        a.desc.set_need_check_feed(False)
         b = -a
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
@@ -266,7 +268,8 @@ class TestMathOpPatches(unittest.TestCase):
 
     @prog_scope()
     def test_astype(self):
-        a = fluid.layers.data(name="a", shape=[10, 1])
+        a = paddle.static.data(name="a", shape=[-1, 10, 1])
+        a.desc.set_need_check_feed(False)
         b = a.astype('float32')
         place = fluid.CPUPlace()
         exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/test_mean_op.py b/python/paddle/fluid/tests/unittests/test_mean_op.py
index c6ade1ca53d0f0515acec4aea1dccb623f8373c8..3fcd423aa56c532128c9f43e90bb9d97af102c52 100644
--- a/python/paddle/fluid/tests/unittests/test_mean_op.py
+++ b/python/paddle/fluid/tests/unittests/test_mean_op.py
@@ -23,7 +23,6 @@ from test_sum_op import TestReduceOPTensorAxisBase
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 
 np.random.seed(10)
@@ -82,12 +81,12 @@ class TestMeanOpError(unittest.TestCase):
             input1 = 12
             self.assertRaises(TypeError, paddle.mean, input1)
             # The input dtype of mean_op must be float16, float32, float64.
-            input2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            input2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.mean, input2)
-            input3 = fluid.layers.data(
-                name='input3', shape=[4], dtype="float16"
+            input3 = paddle.static.data(
+                name='input3', shape=[-1, 4], dtype="float16"
             )
             paddle.nn.functional.softmax(input3)
 
@@ -442,7 +441,7 @@ class TestMeanDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [3, 4, 5], False, dtype)
+        data = paddle.static.data('data', [3, 4, 5], dtype)
         data.persistable = True
         out = paddle.mean(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -473,7 +472,7 @@ class TestMeanTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [3, 4, 5], False, dtype)
+        data = paddle.static.data('data', [3, 4, 5], dtype)
         data.persistable = True
         out = paddle.mean(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_memory_reuse_exclude_feed_var.py b/python/paddle/fluid/tests/unittests/test_memory_reuse_exclude_feed_var.py
index bdd6b3d30489937fcb5fe32667c65c5fc635c70a..32ce652ace45ab6ecf774cfa20fc716678055bf6 100644
--- a/python/paddle/fluid/tests/unittests/test_memory_reuse_exclude_feed_var.py
+++ b/python/paddle/fluid/tests/unittests/test_memory_reuse_exclude_feed_var.py
@@ -27,8 +27,8 @@ class TestMemoryReuseExcludeFeedVar(unittest.TestCase):
         self.iteration = 10
 
     def main_impl(self, place):
-        image = fluid.layers.data(
-            name='image', shape=self.image_shape, dtype='float32'
+        image = paddle.static.data(
+            name='image', shape=[-1] + self.image_shape, dtype='float32'
         )
         relu_image = F.relu(image)
         loss = paddle.mean(relu_image)
diff --git a/python/paddle/fluid/tests/unittests/test_memory_usage.py b/python/paddle/fluid/tests/unittests/test_memory_usage.py
index b083bc6d3051b80e5cf0053a76e062dae785d2a1..a9b2d2a92b86fa8fe14844cd86388e852cc507aa 100644
--- a/python/paddle/fluid/tests/unittests/test_memory_usage.py
+++ b/python/paddle/fluid/tests/unittests/test_memory_usage.py
@@ -26,9 +26,9 @@ def train_simulator(test_batch_size=10):
             "but got batch_size={}".format(test_batch_size)
         )
 
-    x = fluid.layers.data(name='x', shape=[13], dtype='float32')
+    x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
     y_predict = paddle.static.nn.fc(x, size=1, activation=None)
-    y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+    y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
 
     cost = paddle.nn.functional.square_error_cost(input=y_predict, label=y)
     avg_cost = paddle.mean(cost)
diff --git a/python/paddle/fluid/tests/unittests/test_mix_precision_all_reduce_fuse.py b/python/paddle/fluid/tests/unittests/test_mix_precision_all_reduce_fuse.py
index 20bf8e0985aa3cc3b2ed3f002af864024dc91d4d..6a1c6af11fd026e45eb479f09022912984ae5aad 100644
--- a/python/paddle/fluid/tests/unittests/test_mix_precision_all_reduce_fuse.py
+++ b/python/paddle/fluid/tests/unittests/test_mix_precision_all_reduce_fuse.py
@@ -36,8 +36,10 @@ def loss_net(hidden, label):
 
 
 def conv_net(use_feed):
-    img = fluid.layers.data(name='image', shape=img_shape, dtype='float16')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(
+        name='image', shape=[-1] + img_shape, dtype='float16'
+    )
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     conv_pool_1 = fluid.nets.simple_img_conv_pool(
         input=img,
diff --git a/python/paddle/fluid/tests/unittests/test_momentum_op.py b/python/paddle/fluid/tests/unittests/test_momentum_op.py
index 4b745cb7a6fb0c682332b04811a26612fc3566e8..6ec02735d37d258eecbfa3b5d248a2fe3b8689b0 100644
--- a/python/paddle/fluid/tests/unittests/test_momentum_op.py
+++ b/python/paddle/fluid/tests/unittests/test_momentum_op.py
@@ -528,8 +528,8 @@ class TestMomentumV2(unittest.TestCase):
         place = fluid.CPUPlace()
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
@@ -666,8 +666,8 @@ class TestMomentumOpWithDecayAPI(unittest.TestCase):
         place = fluid.CPUPlace()
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
diff --git a/python/paddle/fluid/tests/unittests/test_monitor.py b/python/paddle/fluid/tests/unittests/test_monitor.py
index 205414668f9f0cdd2048d1aceade78e594c54d12..a5d5e30176fb07753a1ec9e47c4031fccf825b92 100644
--- a/python/paddle/fluid/tests/unittests/test_monitor.py
+++ b/python/paddle/fluid/tests/unittests/test_monitor.py
@@ -54,8 +54,8 @@ class TestDatasetWithStat(unittest.TestCase):
         slots = ["slot1", "slot2", "slot3", "slot4"]
         slots_vars = []
         for slot in slots:
-            var = fluid.layers.data(
-                name=slot, shape=[1], dtype="int64", lod_level=1
+            var = paddle.static.data(
+                name=slot, shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots_vars.append(var)
 
diff --git a/python/paddle/fluid/tests/unittests/test_mse_loss.py b/python/paddle/fluid/tests/unittests/test_mse_loss.py
index 99fbe012953f3dbd35696c7927df45deb155d5c7..75dafb1ea3553a8cadf7127427956194cdffd285 100644
--- a/python/paddle/fluid/tests/unittests/test_mse_loss.py
+++ b/python/paddle/fluid/tests/unittests/test_mse_loss.py
@@ -79,12 +79,14 @@ class TestNNMseLoss(unittest.TestCase):
                 else fluid.CPUPlace()
             )
             with fluid.program_guard(prog, startup_prog):
-                input = fluid.layers.data(
-                    name='input', shape=dim, dtype='float32'
+                input = paddle.static.data(
+                    name='input', shape=[-1] + dim, dtype='float32'
                 )
-                label = fluid.layers.data(
-                    name='label', shape=dim, dtype='float32'
+                input.desc.set_need_check_feed(False)
+                label = paddle.static.data(
+                    name='label', shape=[-1] + dim, dtype='float32'
                 )
+                label.desc.set_need_check_feed(False)
                 mse_loss = paddle.nn.loss.MSELoss()
                 ret = mse_loss(input, label)
 
@@ -123,12 +125,14 @@ class TestNNMseLoss(unittest.TestCase):
                 else fluid.CPUPlace()
             )
             with fluid.program_guard(prog, startup_prog):
-                input = fluid.layers.data(
-                    name='input', shape=dim, dtype='float32'
+                input = paddle.static.data(
+                    name='input', shape=[-1] + dim, dtype='float32'
                 )
-                label = fluid.layers.data(
-                    name='label', shape=dim, dtype='float32'
+                input.desc.set_need_check_feed(False)
+                label = paddle.static.data(
+                    name='label', shape=[-1] + dim, dtype='float32'
                 )
+                label.desc.set_need_check_feed(False)
                 mse_loss = paddle.nn.loss.MSELoss(reduction='sum')
                 ret = mse_loss(input, label)
 
@@ -167,12 +171,14 @@ class TestNNMseLoss(unittest.TestCase):
                 else fluid.CPUPlace()
             )
             with fluid.program_guard(prog, startup_prog):
-                input = fluid.layers.data(
-                    name='input', shape=dim, dtype='float32'
+                input = paddle.static.data(
+                    name='input', shape=[-1] + dim, dtype='float32'
                 )
-                label = fluid.layers.data(
-                    name='label', shape=dim, dtype='float32'
+                input.desc.set_need_check_feed(False)
+                label = paddle.static.data(
+                    name='label', shape=[-1] + dim, dtype='float32'
                 )
+                label.desc.set_need_check_feed(False)
                 mse_loss = paddle.nn.loss.MSELoss(reduction='none')
                 ret = mse_loss(input, label)
 
diff --git a/python/paddle/fluid/tests/unittests/test_multihead_attention.py b/python/paddle/fluid/tests/unittests/test_multihead_attention.py
index 9d94a7713b1f6e8dc05ddc06c12a5a9391bdcd1e..2db080edbfd9b35d412ef7c0c3d829ce5b04f877 100644
--- a/python/paddle/fluid/tests/unittests/test_multihead_attention.py
+++ b/python/paddle/fluid/tests/unittests/test_multihead_attention.py
@@ -31,18 +31,16 @@ class TestMultiheadAttention(unittest.TestCase):
 
     def set_program(self):
         """Build the test program."""
-        queries = fluid.layers.data(
+        queries = paddle.static.data(
             name="queries",
             shape=self.input_shape,
             dtype="float32",
-            append_batch_size=False,
         )
         queries.stop_gradient = False
-        keys = fluid.layers.data(
+        keys = paddle.static.data(
             name="keys",
             shape=self.input_shape,
             dtype="float32",
-            append_batch_size=False,
         )
         keys.stop_gradient = False
 
diff --git a/python/paddle/fluid/tests/unittests/test_name_scope.py b/python/paddle/fluid/tests/unittests/test_name_scope.py
index 372547722cb1878c331a441e95ccd51e62c2486e..daecd21ab7af819ca9a6c7f0f055299a1e3e00d5 100644
--- a/python/paddle/fluid/tests/unittests/test_name_scope.py
+++ b/python/paddle/fluid/tests/unittests/test_name_scope.py
@@ -21,7 +21,7 @@ import paddle.fluid as fluid
 class TestNameScope(unittest.TestCase):
     def test_name_scope(self):
         with fluid.name_scope("s1"):
-            a = fluid.layers.data(name='data', shape=[1], dtype='int32')
+            a = paddle.static.data(name='data', shape=[-1, 1], dtype='int32')
             b = a + 1
             with fluid.name_scope("s2"):
                 c = b * 1
diff --git a/python/paddle/fluid/tests/unittests/test_nce.py b/python/paddle/fluid/tests/unittests/test_nce.py
index ee51b0d608ad3fa283e14dcdfaf4d74efad49163..e2923da7113df8035429bad5c4cffedbf85ccb74 100644
--- a/python/paddle/fluid/tests/unittests/test_nce.py
+++ b/python/paddle/fluid/tests/unittests/test_nce.py
@@ -187,8 +187,10 @@ class TestNCECase1SelectedRows(unittest.TestCase):
         custom_dist,
         is_sparse,
     ):
-        input = fluid.layers.data(name="input", shape=[10], dtype="float32")
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        input = paddle.static.data(
+            name="input", shape=[-1, 10], dtype="float32"
+        )
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
 
         w_param = (
             fluid.default_main_program()
@@ -288,7 +290,7 @@ class TestNCE_OpError(unittest.TestCase):
             input1 = fluid.create_lod_tensor(
                 np.array([0.0, 3.0, 2.0, 4.0]), [[1, 1, 2]], fluid.CPUPlace()
             )
-            label1 = fluid.layers.data(
+            label1 = paddle.static.data(
                 name='label1', shape=[-1, 4], dtype="int64"
             )
             # the input(input) of nce layer must be Variable.
@@ -296,7 +298,7 @@ class TestNCE_OpError(unittest.TestCase):
                 TypeError, paddle.static.nn.nce, input1, label1, 5
             )
 
-            input2 = fluid.layers.data(
+            input2 = paddle.static.data(
                 name='input2', shape=[-1, 4], dtype="float32"
             )
             label2 = fluid.create_lod_tensor(
@@ -307,10 +309,10 @@ class TestNCE_OpError(unittest.TestCase):
                 TypeError, paddle.static.nn.nce, input2, label2, 5
             )
 
-            input3 = fluid.layers.data(
+            input3 = paddle.static.data(
                 name='input3', shape=[-1, 4], dtype="float16"
             )
-            label3 = fluid.layers.data(
+            label3 = paddle.static.data(
                 name='label3', shape=[-1, 1], dtype="int64"
             )
             # the data type of input(input) must be float32 or float64.
@@ -318,10 +320,10 @@ class TestNCE_OpError(unittest.TestCase):
                 TypeError, paddle.static.nn.nce, input3, label3, 5
             )
 
-            input4 = fluid.layers.data(
+            input4 = paddle.static.data(
                 name='input4', shape=[-1, 4], dtype="float32"
             )
-            label4 = fluid.layers.data(
+            label4 = paddle.static.data(
                 name='label4', shape=[-1, 1], dtype="int32"
             )
             # the data type of input(label) must be int64.
diff --git a/python/paddle/fluid/tests/unittests/test_network_with_dtype.py b/python/paddle/fluid/tests/unittests/test_network_with_dtype.py
index cdde0c1a468ec6960377dc8a448f919fb020428e..19aa2b72198b9bbd393bd34397aff4b43e09e0bf 100644
--- a/python/paddle/fluid/tests/unittests/test_network_with_dtype.py
+++ b/python/paddle/fluid/tests/unittests/test_network_with_dtype.py
@@ -30,8 +30,8 @@ class TestNetWithDtype(unittest.TestCase):
         main = fluid.Program()
         startup = fluid.Program()
         with fluid.program_guard(main, startup):
-            x = fluid.layers.data(name='x', shape=[13], dtype=self.dtype)
-            y = fluid.layers.data(name='y', shape=[1], dtype=self.dtype)
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype=self.dtype)
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype=self.dtype)
             y_predict = paddle.static.nn.fc(x, size=1, activation=None)
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
diff --git a/python/paddle/fluid/tests/unittests/test_nn_functional_embedding_static.py b/python/paddle/fluid/tests/unittests/test_nn_functional_embedding_static.py
index 1051ef8d4b00db5c261dd2c1def58a13b983b9c7..d89af631baa45b8829e9f38618acb1229aef02ba 100644
--- a/python/paddle/fluid/tests/unittests/test_nn_functional_embedding_static.py
+++ b/python/paddle/fluid/tests/unittests/test_nn_functional_embedding_static.py
@@ -16,6 +16,7 @@ import unittest
 
 import numpy as np
 
+import paddle
 import paddle.fluid as fluid
 import paddle.nn.functional as functional
 
@@ -41,10 +42,9 @@ class EmbeddingStatic(unittest.TestCase):
                     (128, 100), attr=param_attr, dtype="float32"
                 )
 
-                label = fluid.layers.data(
+                label = paddle.static.data(
                     name="label",
-                    shape=[4],
-                    append_batch_size=False,
+                    shape=[-1, 4],
                     dtype="int64",
                 )
 
@@ -74,10 +74,9 @@ class EmbeddingStatic(unittest.TestCase):
                     (128, 100), attr=param_attr, dtype="float32"
                 )
 
-                label = fluid.layers.data(
+                label = paddle.static.data(
                     name="label",
-                    shape=[4],
-                    append_batch_size=False,
+                    shape=[-1, 4],
                     dtype="int32",
                 )
 
diff --git a/python/paddle/fluid/tests/unittests/test_nn_functional_hot_op.py b/python/paddle/fluid/tests/unittests/test_nn_functional_hot_op.py
index 0faf226fac4d4f198b958405964836da0bf8e992..0043f15aed3a8c15d88aaa93b03657e4e2c2363c 100644
--- a/python/paddle/fluid/tests/unittests/test_nn_functional_hot_op.py
+++ b/python/paddle/fluid/tests/unittests/test_nn_functional_hot_op.py
@@ -134,9 +134,13 @@ class TestOneHotOp_exception(unittest.TestCase):
     def test_check_output(self):
         program = Program()
         with program_guard(program):
-            x = fluid.layers.data(
-                name='x', shape=[self.dimension], dtype='float32', lod_level=1
+            x = paddle.static.data(
+                name='x',
+                shape=[-1, self.dimension],
+                dtype='float32',
+                lod_level=1,
             )
+            x.desc.set_need_check_feed(False)
             block = program.current_block()
             one_hot_out = block.create_var(
                 name="one_hot_out",
@@ -181,7 +185,8 @@ class TestOneHotOpApi(unittest.TestCase):
             )
 
     def _run(self, num_classes):
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
+        label.desc.set_need_check_feed(False)
         one_hot_label = functional.one_hot(x=label, num_classes=num_classes)
 
         place = fluid.CPUPlace()
@@ -205,12 +210,12 @@ class BadInputTestOnehotV2(unittest.TestCase):
         with fluid.program_guard(fluid.Program()):
 
             def test_bad_x():
-                label = fluid.layers.data(
+                label = paddle.static.data(
                     name="label",
                     shape=[4],
-                    append_batch_size=False,
                     dtype="float32",
                 )
+                label.desc.set_need_check_feed(False)
                 one_hot_label = functional.one_hot(x=label, num_classes=4)
 
             self.assertRaises(TypeError, test_bad_x)
diff --git a/python/paddle/fluid/tests/unittests/test_nn_grad.py b/python/paddle/fluid/tests/unittests/test_nn_grad.py
index ec355baefa9f67d96312c9d87da62086da8f1a8e..b86d3029ddd5a35ec1c4d50516b031bf4a2b224b 100644
--- a/python/paddle/fluid/tests/unittests/test_nn_grad.py
+++ b/python/paddle/fluid/tests/unittests/test_nn_grad.py
@@ -21,7 +21,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -73,7 +72,7 @@ class TestReduceMeanWithDimDoubleGradCheck(unittest.TestCase):
         eps = 0.05
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
         y = paddle.mean(x, axis=0)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -97,7 +96,7 @@ class TestReduceSumWithDimDoubleGradCheck(unittest.TestCase):
         eps = 0.05
         dtype = np.float64
 
-        x = layers.data('x', shape, False, dtype)
+        x = paddle.static.data('x', shape, dtype)
         x.persistable = True
         y = paddle.sum(x, axis=0)
         x_arr = np.random.uniform(-1, 1, shape).astype(dtype)
@@ -122,7 +121,7 @@ class TestReshapeDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.reshape(x, new_shape)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -150,7 +149,7 @@ class TestTileDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.tile(x, repeat_times)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -181,7 +180,7 @@ class TestExpandV2DoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.expand(x, new_shape)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -213,7 +212,7 @@ class TestSqueezeDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.squeeze(x, axes)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -245,7 +244,7 @@ class TestUnsqueezeDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.unsqueeze(x, axes)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -274,7 +273,7 @@ class TestClipDoubleGradCheck(unittest.TestCase):
         x_shape = [2, 4, 10]
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.clip(x, min=-1.0, max=1.0)
         x_arr = np.random.uniform(-5.0, 5.0, x_shape).astype(dtype)
@@ -299,7 +298,7 @@ class TestTransposeDoubleGradCheck(unittest.TestCase):
         perm = [1, 0]
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.transpose(x, perm)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -321,7 +320,7 @@ class TestTransposeDoubleGradCheckCase1(unittest.TestCase):
         perm = [0, 2, 3, 1]
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.transpose(x, perm)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -348,7 +347,7 @@ class TestConstantPadDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.nn.functional.pad(x, pad)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -375,7 +374,7 @@ class TestConstantPadDoubleGradCheckCase1(TestConstantPadDoubleGradCheck):
         pad = [1, 0, 1, 0, 1, 0, 1, 0]
         dtype = np.float64
 
-        x = layers.data('x', x_shape, False, dtype)
+        x = paddle.static.data('x', x_shape, dtype)
         x.persistable = True
         out = paddle.nn.functional.pad(x, pad)
         x_arr = np.random.uniform(-1, 1, x_shape).astype(dtype)
@@ -393,8 +392,8 @@ class TestConcatDoubleGradCheck(unittest.TestCase):
         pad = [1, 1, 1, 1]
         dtype = np.float64
 
-        x1 = layers.data('x', x_shape, False, dtype)
-        x2 = layers.data('x', x_shape, False, dtype)
+        x1 = paddle.static.data('x', x_shape, dtype)
+        x2 = paddle.static.data('x', x_shape, dtype)
         x1.persistable = True
         x2.persistable = True
         out = paddle.concat([x1, x2], axis=0)
@@ -423,10 +422,9 @@ class TestConcatDoubleGradCheck(unittest.TestCase):
 class TestAvgPool2DDoubleGradCheckCase1(unittest.TestCase):
     @prog_scope()
     def func(self, place):
-        input_NCHW = fluid.layers.data(
+        input_NCHW = paddle.static.data(
             name="input_NCHW",
             shape=[2, 3, 5, 5],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -454,10 +452,9 @@ class TestAvgPool2DDoubleGradCheckCase2(unittest.TestCase):
 
     @prog_scope()
     def func(self, place):
-        input_NHWC = fluid.layers.data(
+        input_NHWC = paddle.static.data(
             name="input_NHWC",
             shape=[2, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -491,10 +488,9 @@ class TestAvgPool2DDoubleGradCheckCase3(unittest.TestCase):
 
     @prog_scope()
     def func(self, place):
-        input_NCHW = fluid.layers.data(
+        input_NCHW = paddle.static.data(
             name="input_NCHW",
             shape=[2, 3, 5, 5],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -525,10 +521,9 @@ class TestAvgPool2DDoubleGradCheckCase4(unittest.TestCase):
 
     @prog_scope()
     def func(self, place):
-        input_NCHW = fluid.layers.data(
+        input_NCHW = paddle.static.data(
             name="input_NCHW",
             shape=[2, 3, 5, 5],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/test_nonzero_api.py b/python/paddle/fluid/tests/unittests/test_nonzero_api.py
index dce29b96e5559653cc72ddbfcfee35ca35bfc474..aac65fba54c8526569881ae1b8c019116a8074c9 100644
--- a/python/paddle/fluid/tests/unittests/test_nonzero_api.py
+++ b/python/paddle/fluid/tests/unittests/test_nonzero_api.py
@@ -25,7 +25,8 @@ class TestNonZeroAPI(unittest.TestCase):
     def test_nonzero_api_as_tuple(self):
         data = np.array([[True, False], [False, True]])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 2])
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = paddle.nonzero(x, as_tuple=True)
             self.assertEqual(type(y), tuple)
             self.assertEqual(len(y), 2)
@@ -40,7 +41,8 @@ class TestNonZeroAPI(unittest.TestCase):
 
         data = np.array([True, True, False])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1])
+            x = paddle.static.data(name='x', shape=[-1], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = paddle.nonzero(x, as_tuple=True)
             self.assertEqual(type(y), tuple)
             self.assertEqual(len(y), 1)
@@ -55,7 +57,8 @@ class TestNonZeroAPI(unittest.TestCase):
     def test_nonzero_api(self):
         data = np.array([[True, False], [False, True]])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 2])
+            x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = paddle.nonzero(x)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -66,7 +69,8 @@ class TestNonZeroAPI(unittest.TestCase):
 
         data = np.array([True, True, False])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1])
+            x = paddle.static.data(name='x', shape=[-1], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = paddle.nonzero(x)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
diff --git a/python/paddle/fluid/tests/unittests/test_normalization_wrapper.py b/python/paddle/fluid/tests/unittests/test_normalization_wrapper.py
index 60cc8c7665edb9536d012bcb179fc1b20e01c60c..8ef60fce49806c3bf9c7c44c09f2cc2ec508ef5b 100644
--- a/python/paddle/fluid/tests/unittests/test_normalization_wrapper.py
+++ b/python/paddle/fluid/tests/unittests/test_normalization_wrapper.py
@@ -32,11 +32,10 @@ class TestNormalization(unittest.TestCase):
 
     def set_program(self, axis, epsilon):
         """Build the test program."""
-        data = fluid.layers.data(
+        data = paddle.static.data(
             name=self.data_desc["name"],
             shape=self.data_desc["shape"],
             dtype="float32",
-            append_batch_size=False,
         )
         data.stop_gradient = False
         l2_norm = paddle.nn.functional.normalize(
diff --git a/python/paddle/fluid/tests/unittests/test_npair_loss_op.py b/python/paddle/fluid/tests/unittests/test_npair_loss_op.py
index efbf8a98a87b8d761d4f0962162521286889f4a3..841d7acc2c0864cdf703cd34b8e43896e168584a 100644
--- a/python/paddle/fluid/tests/unittests/test_npair_loss_op.py
+++ b/python/paddle/fluid/tests/unittests/test_npair_loss_op.py
@@ -83,23 +83,20 @@ class TestNpairLossOp(unittest.TestCase):
             l2_reg=reg_lambda,
         )
 
-        anc = fluid.layers.data(
+        anc = paddle.static.data(
             dtype='float32',
             name='anc',
             shape=embeddings_anchor.shape,
-            append_batch_size=False,
         )
-        pos = fluid.layers.data(
+        pos = paddle.static.data(
             dtype='float32',
             name='pos',
             shape=embeddings_positive.shape,
-            append_batch_size=False,
         )
-        lab = fluid.layers.data(
+        lab = paddle.static.data(
             dtype='float32',
             name='lab',
             shape=row_labels.shape,
-            append_batch_size=False,
         )
 
         npair_loss_op = paddle.nn.functional.npair_loss(
diff --git a/python/paddle/fluid/tests/unittests/test_one_hot_v2_op.py b/python/paddle/fluid/tests/unittests/test_one_hot_v2_op.py
index 53e34ae70e6739c8c6228bd73f1e02e777a7c6fc..30bb75e0fa7833ef213a9c9547fea0761aab940c 100644
--- a/python/paddle/fluid/tests/unittests/test_one_hot_v2_op.py
+++ b/python/paddle/fluid/tests/unittests/test_one_hot_v2_op.py
@@ -133,9 +133,13 @@ class TestOneHotOp_exception(unittest.TestCase):
     def test_check_output(self):
         program = Program()
         with program_guard(program):
-            x = fluid.layers.data(
-                name='x', shape=[self.dimension], dtype='float32', lod_level=1
+            x = paddle.static.data(
+                name='x',
+                shape=[-1, self.dimension],
+                dtype='float32',
+                lod_level=1,
             )
+            x.desc.set_need_check_feed(False)
             block = program.current_block()
             one_hot_out = block.create_var(
                 name="one_hot_out",
@@ -187,7 +191,8 @@ class TestOneHotOpApi(unittest.TestCase):
             )
 
     def _run(self, depth):
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
+        label.desc.set_need_check_feed(False)
         one_hot_label = fluid.one_hot(input=label, depth=depth)
 
         place = fluid.CPUPlace()
@@ -211,12 +216,12 @@ class BadInputTestOnehotV2(unittest.TestCase):
         with fluid.program_guard(fluid.Program()):
 
             def test_bad_x():
-                label = fluid.layers.data(
+                label = paddle.static.data(
                     name="label",
-                    shape=[4],
-                    append_batch_size=False,
+                    shape=[-1, 4],
                     dtype="float32",
                 )
+                label.desc.set_need_check_feed(False)
                 one_hot_label = fluid.one_hot(input=label, depth=4)
 
             self.assertRaises(TypeError, test_bad_x)
diff --git a/python/paddle/fluid/tests/unittests/test_optimizer.py b/python/paddle/fluid/tests/unittests/test_optimizer.py
index d5f54d44829525474fdc9e365435d091e4ddfcca..20a4a7705e9aecacacbd413d42403f02d68626dc 100644
--- a/python/paddle/fluid/tests/unittests/test_optimizer.py
+++ b/python/paddle/fluid/tests/unittests/test_optimizer.py
@@ -1181,10 +1181,12 @@ class TestRecomputeOptimizer(unittest.TestCase):
         scope = fluid.Scope()
         with fluid.scope_guard(scope):
             with program_guard(main_program, startup_program):
-                input_x = fluid.layers.data(
-                    name="x", shape=[3], dtype='float32'
+                input_x = paddle.static.data(
+                    name="x", shape=[-1, 3], dtype='float32'
+                )
+                input_y = paddle.static.data(
+                    name="y", shape=[-1, 1], dtype='int64'
                 )
-                input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
                 drop_res, prediction, cost = mlp(input_x, input_y)
                 sgd = fluid.optimizer.Adam(learning_rate=0.01)
                 sgd = fluid.optimizer.RecomputeOptimizer(sgd)
@@ -1243,10 +1245,12 @@ class TestRecomputeOptimizerCUDA(unittest.TestCase):
         scope = fluid.Scope()
         with fluid.scope_guard(scope):
             with program_guard(main_program, startup_program):
-                input_x = fluid.layers.data(
-                    name="x", shape=[3], dtype='float32'
+                input_x = paddle.static.data(
+                    name="x", shape=[-1, 3], dtype='float32'
+                )
+                input_y = paddle.static.data(
+                    name="y", shape=[-1, 1], dtype='int64'
                 )
-                input_y = fluid.layers.data(name="y", shape=[1], dtype='int64')
                 drop_res, prediction, cost = mlp(input_x, input_y)
                 sgd = fluid.optimizer.Adam(learning_rate=0.01)
                 sgd = fluid.optimizer.RecomputeOptimizer(sgd)
diff --git a/python/paddle/fluid/tests/unittests/test_optimizer_in_control_flow.py b/python/paddle/fluid/tests/unittests/test_optimizer_in_control_flow.py
index 731693f5cf7fc47f996236349bc6fb0f670b412f..3b32c9ca4ee7886dac1aa0f90786227b2d57507e 100644
--- a/python/paddle/fluid/tests/unittests/test_optimizer_in_control_flow.py
+++ b/python/paddle/fluid/tests/unittests/test_optimizer_in_control_flow.py
@@ -264,7 +264,7 @@ class TestMultiOptimizersMultiCardsError(unittest.TestCase):
             def fn_2(opt, avg_loss):
                 opt.minimize(avg_loss)
 
-            x = fluid.layers.data("X", [10], 'float32')
+            x = paddle.static.data("X", [-1, 10], 'float32')
             hidden = paddle.static.nn.fc(x, 5)
             avg_loss = paddle.mean(hidden)
 
diff --git a/python/paddle/fluid/tests/unittests/test_parallel_executor_drop_scope.py b/python/paddle/fluid/tests/unittests/test_parallel_executor_drop_scope.py
index 8654f0ba5e97c0e86c1242003191bd80045f28b9..0984b02d750678c6ba904527a19f2edcf164a2d6 100644
--- a/python/paddle/fluid/tests/unittests/test_parallel_executor_drop_scope.py
+++ b/python/paddle/fluid/tests/unittests/test_parallel_executor_drop_scope.py
@@ -31,7 +31,7 @@ class TestParallelExecutorDropExeScope(unittest.TestCase):
         train_program = fluid.Program()
         startup_program = fluid.Program()
         with fluid.program_guard(train_program, startup_program):
-            data = fluid.layers.data(name='X', shape=[1], dtype='float32')
+            data = paddle.static.data(name='X', shape=[-1, 1], dtype='float32')
             hidden = paddle.static.nn.fc(x=data, size=10)
             loss = paddle.mean(hidden)
             test_program = fluid.default_main_program().clone(for_test=True)
diff --git a/python/paddle/fluid/tests/unittests/test_parallel_executor_dry_run.py b/python/paddle/fluid/tests/unittests/test_parallel_executor_dry_run.py
index 93fecfefb5a4475f24dc941bff7a3ae12cf39b2b..0d750ddcbe4c2660f303f6d262fd9e493ac242c1 100644
--- a/python/paddle/fluid/tests/unittests/test_parallel_executor_dry_run.py
+++ b/python/paddle/fluid/tests/unittests/test_parallel_executor_dry_run.py
@@ -76,8 +76,8 @@ class TestMNISTDryRun(TestBase):
 
     @staticmethod
     def network_func():
-        img = fluid.layers.data(name='img', shape=[784], dtype='float32')
-        label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+        img = paddle.static.data(name='img', shape=[-1, 784], dtype='float32')
+        label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
         hidden = img
         for _ in range(10):
             hidden = paddle.static.nn.fc(x=img, size=200, activation='tanh')
diff --git a/python/paddle/fluid/tests/unittests/test_parallel_executor_fetch_feed.py b/python/paddle/fluid/tests/unittests/test_parallel_executor_fetch_feed.py
index c92d3234597f7d73cb39fa533c242c5b8421a526..5aa87bc7b06982d85115836d873e8c819a89e146 100644
--- a/python/paddle/fluid/tests/unittests/test_parallel_executor_fetch_feed.py
+++ b/python/paddle/fluid/tests/unittests/test_parallel_executor_fetch_feed.py
@@ -55,10 +55,12 @@ class TestFetchAndFeed(unittest.TestCase):
         startup = fluid.Program()
         startup.random_seed = seed
         with fluid.program_guard(main_program, startup):
-            data = fluid.layers.data(
-                name='image', shape=[3, 224, 224], dtype='float32'
+            data = paddle.static.data(
+                name='image', shape=[-1, 3, 224, 224], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             out = Lenet(data, class_dim=102)
             loss = paddle.nn.functional.cross_entropy(
                 input=out, label=label, reduction='none', use_softmax=False
diff --git a/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py b/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py
index d81586fe251c34aa319b91759e5136550e1dd583..850ddc379c6092738c1deb7197d9a130258b52c9 100644
--- a/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py
+++ b/python/paddle/fluid/tests/unittests/test_parallel_executor_mnist.py
@@ -24,8 +24,8 @@ import paddle.fluid.core as core
 
 
 def simple_fc_net(use_feed):
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     hidden = img
     for _ in range(4):
         hidden = paddle.static.nn.fc(
@@ -45,8 +45,8 @@ def simple_fc_net(use_feed):
 
 
 def fc_with_batchnorm(use_feed):
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     hidden = img
     for _ in range(1):
diff --git a/python/paddle/fluid/tests/unittests/test_print_op.py b/python/paddle/fluid/tests/unittests/test_print_op.py
index 75e8e93987dd5c77bc1af8ab0f7d5b33976cbaa1..19be1c7526701f912f96efaab84b3b979652439e 100755
--- a/python/paddle/fluid/tests/unittests/test_print_op.py
+++ b/python/paddle/fluid/tests/unittests/test_print_op.py
@@ -19,7 +19,6 @@ from simple_nets import init_data, simple_fc_net
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 from paddle.fluid.framework import switch_main_program
 from paddle.static import Program, program_guard
@@ -36,7 +35,7 @@ class TestPrintOpCPU(unittest.TestCase):
         self.x_tensor.set_recursive_sequence_lengths([[1, 1]])
 
     def build_network(self, only_forward, **kargs):
-        x = layers.data('x', shape=[3], dtype='float32', lod_level=1)
+        x = paddle.static.data('x', shape=[-1, 3], dtype='float32', lod_level=1)
         x.stop_gradient = False
         paddle.static.Print(input=x, **kargs)
         loss = paddle.mean(x)
@@ -60,7 +59,7 @@ class TestPrintOpCPU(unittest.TestCase):
         )
 
     def test_all_parameters(self):
-        x = layers.data('x', shape=[3], dtype='float32', lod_level=1)
+        x = paddle.static.data('x', shape=[-1, 3], dtype='float32', lod_level=1)
         x.stop_gradient = False
 
         for print_tensor_name in [True, False]:
diff --git a/python/paddle/fluid/tests/unittests/test_profiler.py b/python/paddle/fluid/tests/unittests/test_profiler.py
index b20538cad5b6679ad49c2a3f9357bac420b9a812..e22deb5df86df3ba6417178267258deebdce3269 100644
--- a/python/paddle/fluid/tests/unittests/test_profiler.py
+++ b/python/paddle/fluid/tests/unittests/test_profiler.py
@@ -37,7 +37,9 @@ class TestProfiler(unittest.TestCase):
         startup_program = fluid.Program()
         main_program = fluid.Program()
         with fluid.program_guard(main_program, startup_program):
-            image = fluid.layers.data(name='x', shape=[784], dtype='float32')
+            image = paddle.static.data(
+                name='x', shape=[-1, 784], dtype='float32'
+            )
             hidden1 = paddle.static.nn.fc(x=image, size=64, activation='relu')
             i = layers.zeros(shape=[1], dtype='int64')
             counter = fluid.layers.zeros(
@@ -62,7 +64,7 @@ class TestProfiler(unittest.TestCase):
             predict = paddle.static.nn.fc(
                 x=hidden2, size=10, activation='softmax'
             )
-            label = fluid.layers.data(name='y', shape=[1], dtype='int64')
+            label = paddle.static.data(name='y', shape=[-1, 1], dtype='int64')
             cost = paddle.nn.functional.cross_entropy(
                 input=predict, label=label, reduction='none', use_softmax=False
             )
diff --git a/python/paddle/fluid/tests/unittests/test_program.py b/python/paddle/fluid/tests/unittests/test_program.py
index 56b8b35234d8b145ff1c1106345e029e15198573..dfc9b7572da4bf5266403143932db23d950cf25c 100644
--- a/python/paddle/fluid/tests/unittests/test_program.py
+++ b/python/paddle/fluid/tests/unittests/test_program.py
@@ -16,7 +16,6 @@ import unittest
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid.framework import Program, default_main_program, program_guard
 
 paddle.enable_static()
@@ -98,7 +97,7 @@ class TestProgram(unittest.TestCase):
         main_program = Program()
         startup_program = Program()
         with program_guard(main_program, startup_program):
-            d = layers.data(name='x', shape=[784], dtype='float32')
+            d = paddle.static.data(name='x', shape=[-1, 784], dtype='float32')
             hidden = paddle.static.nn.fc(x=d, size=100)
             paddle.static.nn.fc(x=hidden, size=100)
 
diff --git a/python/paddle/fluid/tests/unittests/test_program_prune_backward.py b/python/paddle/fluid/tests/unittests/test_program_prune_backward.py
index 337feb16174cbfd358d2ae87e2d4daf156f8acd2..5364dcaa6e14a3574fa1d776f9d9706c613bc542 100755
--- a/python/paddle/fluid/tests/unittests/test_program_prune_backward.py
+++ b/python/paddle/fluid/tests/unittests/test_program_prune_backward.py
@@ -30,8 +30,8 @@ import paddle.fluid.core as core
 
 
 def simple_fc_net_with_accuracy(use_feed):
-    img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    img = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
 
     hidden = img
     for _ in range(4):
@@ -53,12 +53,12 @@ def simple_fc_net_with_accuracy(use_feed):
 
 
 def cond_net(use_feed=None):
-    x = fluid.layers.data(name="x", shape=[4], dtype='float32')
-    label = fluid.layers.data('label', shape=[1], dtype='int64')
+    x = paddle.static.data(name="x", shape=[-1, 4], dtype='float32')
+    label = paddle.static.data('label', shape=[-1, 1], dtype='int64')
     prediction = paddle.static.nn.fc(x, size=1, activation=None)
 
     def loss1(pred, label):
-        x = fluid.layers.data(name="x", shape=[4], dtype='float32')
+        x = paddle.static.data(name="x", shape=[-1, 4], dtype='float32')
         loss = paddle.nn.functional.cross_entropy(
             input=pred, label=label, reduction='none', use_softmax=False
         )
@@ -82,12 +82,12 @@ def cond_net(use_feed=None):
 
 
 def optimization_in_cond_net(with_optimize=False):
-    x = fluid.layers.data(name="x", shape=[4], dtype='float32')
-    label = fluid.layers.data('label', shape=[1], dtype='int64')
+    x = paddle.static.data(name="x", shape=[-1, 4], dtype='float32')
+    label = paddle.static.data('label', shape=[-1, 1], dtype='int64')
     prediction = paddle.static.nn.fc(x, size=1, activation=None)
 
     def loss1(opt, pred, label, with_optimize):
-        x = fluid.layers.data(name="x", shape=[4], dtype='float32')
+        x = paddle.static.data(name="x", shape=[-1, 4], dtype='float32')
         loss = paddle.nn.functional.cross_entropy(
             input=pred, label=label, reduction='none', use_softmax=False
         )
diff --git a/python/paddle/fluid/tests/unittests/test_program_to_string.py b/python/paddle/fluid/tests/unittests/test_program_to_string.py
index 55f32b687d914726a57b82106619353b8183cd11..04a816a54125f1037b8a4e4dd7309134fbb78da1 100644
--- a/python/paddle/fluid/tests/unittests/test_program_to_string.py
+++ b/python/paddle/fluid/tests/unittests/test_program_to_string.py
@@ -21,9 +21,7 @@ import paddle.fluid as fluid
 class TestProgram(unittest.TestCase):
     def test_program_to_string(self):
         prog = fluid.default_main_program()
-        a = fluid.layers.data(
-            name="X", shape=[2, 3], dtype="float32", append_batch_size=False
-        )
+        a = paddle.static.data(name="X", shape=[2, 3], dtype="float32")
         c = paddle.static.nn.fc(a, size=3)
         prog_string = prog.to_string(throw_on_error=True, with_details=False)
         prog_string_with_details = prog.to_string(
diff --git a/python/paddle/fluid/tests/unittests/test_prune.py b/python/paddle/fluid/tests/unittests/test_prune.py
index 2c97f49aeea573ea043a2a1433f2ee0cbcb72f26..30e3aefe0a738ca2af431aa89ec32a6db4ae5d19 100644
--- a/python/paddle/fluid/tests/unittests/test_prune.py
+++ b/python/paddle/fluid/tests/unittests/test_prune.py
@@ -25,8 +25,10 @@ import paddle.fluid.framework as framework
 
 class TestPrune(unittest.TestCase):
     def net(self):
-        x = fluid.layers.data(name='x', shape=[2], dtype='float32')
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+        x.desc.set_need_check_feed(False)
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
+        label.desc.set_need_check_feed(False)
         y = paddle.static.nn.fc(x=[x], size=2, activation="softmax")
         loss = paddle.nn.functional.cross_entropy(
             input=y, label=label, reduction='none', use_softmax=False
@@ -161,8 +163,10 @@ def _mock_guard(mock):
 
 class TestExecutorRunAutoPrune(unittest.TestCase):
     def net1(self):
-        x = fluid.layers.data(name='x', shape=[2], dtype='float32')
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
+        x.desc.set_need_check_feed(False)
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
+        label.desc.set_need_check_feed(False)
         w_param_attrs = fluid.ParamAttr(
             name="fc_weight",
             learning_rate=0.5,
@@ -185,9 +189,12 @@ class TestExecutorRunAutoPrune(unittest.TestCase):
         return x, y, label, loss1, loss2, w_param_attrs
 
     def net2(self):
-        x1 = fluid.layers.data(name='x1', shape=[2], dtype='float32')
-        x2 = fluid.layers.data(name='x2', shape=[2], dtype='float32')
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        x1 = paddle.static.data(name='x1', shape=[-1, 2], dtype='float32')
+        x1.desc.set_need_check_feed(False)
+        x2 = paddle.static.data(name='x2', shape=[-1, 2], dtype='float32')
+        x2.desc.set_need_check_feed(False)
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
+        label.desc.set_need_check_feed(False)
         w1_param_attrs = fluid.ParamAttr(
             name="fc_weight1",
             learning_rate=0.5,
diff --git a/python/paddle/fluid/tests/unittests/test_pull_gpups_sparse_op.py b/python/paddle/fluid/tests/unittests/test_pull_gpups_sparse_op.py
index b537a31b3aae8792cebc51890788fa4576f07235..1e50e1a08c679cfdb7f070765f5f77dcca9e32b4 100644
--- a/python/paddle/fluid/tests/unittests/test_pull_gpups_sparse_op.py
+++ b/python/paddle/fluid/tests/unittests/test_pull_gpups_sparse_op.py
@@ -32,8 +32,8 @@ class TestPullGpupsSparse(unittest.TestCase):
         slots = []
         with fluid.program_guard(train_program, startup_program):
 
-            l = fluid.layers.data(
-                name='input', shape=[1], dtype="int64", lod_level=1
+            l = paddle.static.data(
+                name='input', shape=[-1, 1], dtype="int64", lod_level=1
             )
             slots.append(l)
             output = _pull_gpups_sparse(
diff --git a/python/paddle/fluid/tests/unittests/test_py_func_op.py b/python/paddle/fluid/tests/unittests/test_py_func_op.py
index 69e363ee71863c6508f03887973fe6830c76c70b..a90e37a4755c23632e9f54a2b2dbdffbca135d6f 100644
--- a/python/paddle/fluid/tests/unittests/test_py_func_op.py
+++ b/python/paddle/fluid/tests/unittests/test_py_func_op.py
@@ -179,8 +179,12 @@ def test_main(use_cuda, use_py_func_op, use_parallel_executor):
         with fluid.scope_guard(fluid.core.Scope()):
             gen = paddle.seed(1)
             np.random.seed(1)
-            img = fluid.layers.data(name='image', shape=[784], dtype='float32')
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+            img = paddle.static.data(
+                name='image', shape=[-1, 784], dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
+            )
             loss = simple_fc_net(img, label, use_py_func_op)
             optimizer = fluid.optimizer.SGD(learning_rate=1e-3)
             optimizer.minimize(loss)
diff --git a/python/paddle/fluid/tests/unittests/test_py_reader_combination.py b/python/paddle/fluid/tests/unittests/test_py_reader_combination.py
index e238eeaf4cb8544a4a6b19589f5d3950c6562d47..db664c55ebee7fd0c35763e5c3aea11d2961b2d1 100644
--- a/python/paddle/fluid/tests/unittests/test_py_reader_combination.py
+++ b/python/paddle/fluid/tests/unittests/test_py_reader_combination.py
@@ -56,10 +56,12 @@ class TestPyReaderCombination(unittest.TestCase):
 
     def main_impl(self, place):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            image = fluid.layers.data(
-                name='image', dtype='float32', shape=[784]
+            image = paddle.static.data(
+                name='image', dtype='float32', shape=[-1, 784]
+            )
+            label = paddle.static.data(
+                name='label', dtype='int64', shape=[-1, 1]
             )
-            label = fluid.layers.data(name='label', dtype='int64', shape=[1])
 
             py_reader1 = fluid.io.PyReader(
                 feed_list=[image, label], capacity=16, iterable=True
diff --git a/python/paddle/fluid/tests/unittests/test_py_reader_return_list.py b/python/paddle/fluid/tests/unittests/test_py_reader_return_list.py
index d18a66b82854cb73c86a40fa48a5c0df74b1152e..ddef600b166667d618330833d31ef934cd9fc580 100644
--- a/python/paddle/fluid/tests/unittests/test_py_reader_return_list.py
+++ b/python/paddle/fluid/tests/unittests/test_py_reader_return_list.py
@@ -38,8 +38,8 @@ class TestPyReader(unittest.TestCase):
 
         for return_list in [True, False]:
             with fluid.program_guard(fluid.Program(), fluid.Program()):
-                image = fluid.layers.data(
-                    name='image', shape=[784, 784], dtype='float32'
+                image = paddle.static.data(
+                    name='image', shape=[-1, 784, 784], dtype='float32'
                 )
                 reader = fluid.io.PyReader(
                     feed_list=[image],
diff --git a/python/paddle/fluid/tests/unittests/test_py_reader_sample_generator.py b/python/paddle/fluid/tests/unittests/test_py_reader_sample_generator.py
index 19cd4f546acbcd395ad3126004163030632a04e6..afc81e0f2a2d837610bf57cf9a28d1df763265ac 100644
--- a/python/paddle/fluid/tests/unittests/test_py_reader_sample_generator.py
+++ b/python/paddle/fluid/tests/unittests/test_py_reader_sample_generator.py
@@ -54,8 +54,10 @@ class TestCaseBase(unittest.TestCase):
         return ret
 
     def run_main(self, reader, use_sample_generator, iterable, drop_last):
-        image = fluid.layers.data(name='image', dtype='float32', shape=[784])
-        label = fluid.layers.data(name='label', dtype='int64', shape=[1])
+        image = paddle.static.data(
+            name='image', dtype='float32', shape=[-1, 784]
+        )
+        label = paddle.static.data(name='label', dtype='int64', shape=[-1, 1])
         py_reader = fluid.io.PyReader(
             feed_list=[image, label],
             capacity=16,
diff --git a/python/paddle/fluid/tests/unittests/test_reader_reset.py b/python/paddle/fluid/tests/unittests/test_reader_reset.py
index bbdade712e111372db1a008f8032710aa357ca76..9524068316bd696b929d3b175059d609ac9317a1 100644
--- a/python/paddle/fluid/tests/unittests/test_reader_reset.py
+++ b/python/paddle/fluid/tests/unittests/test_reader_reset.py
@@ -46,10 +46,12 @@ class TestReaderReset(unittest.TestCase):
         startup_prog = fluid.Program()
 
         with fluid.program_guard(main_prog, startup_prog):
-            image = fluid.layers.data(
-                name='image', shape=self.ins_shape, dtype='float32'
+            image = paddle.static.data(
+                name='image', shape=[-1] + self.ins_shape, dtype='float32'
+            )
+            label = paddle.static.data(
+                name='label', shape=[-1, 1], dtype='int64'
             )
-            label = fluid.layers.data(name='label', shape=[1], dtype='int64')
             data_reader_handle = fluid.io.PyReader(
                 feed_list=[image, label],
                 capacity=16,
diff --git a/python/paddle/fluid/tests/unittests/test_recurrent_op.py b/python/paddle/fluid/tests/unittests/test_recurrent_op.py
index db1abc9ef0ec441b1448a7fe28242d11c7225c54..4ea5ed0e0d35fdce9b11f7e3df5402d172c67ea4 100644
--- a/python/paddle/fluid/tests/unittests/test_recurrent_op.py
+++ b/python/paddle/fluid/tests/unittests/test_recurrent_op.py
@@ -135,15 +135,14 @@ class RecurrentOpTest1(unittest.TestCase):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot = layers.data(
-            shape=[self.input_dim], dtype='float32', name='h_boot'
+        h_boot = paddle.static.data(
+            shape=[-1, self.input_dim], dtype='float32', name='h_boot'
         )
         h_boot.stop_gradient = False
 
@@ -281,15 +280,14 @@ class RecurrentOpTest2(RecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot = layers.data(
-            shape=[self.input_dim], dtype='float32', name='h_boot'
+        h_boot = paddle.static.data(
+            shape=[-1, self.input_dim], dtype='float32', name='h_boot'
         )
         h_boot.stop_gradient = False
 
@@ -390,25 +388,22 @@ class RecurrentOpMultipleMemoryTest(RecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot1 = layers.data(
+        h_boot1 = paddle.static.data(
             shape=[self.batch_size, self.input_dim],
             dtype='float32',
             name='h_boot1',
-            append_batch_size=False,
         )
         h_boot1.stop_gradient = False
-        h_boot2 = layers.data(
+        h_boot2 = paddle.static.data(
             shape=[self.batch_size, self.input_dim],
             dtype='float32',
             name='h_boot2',
-            append_batch_size=False,
         )
         h_boot2.stop_gradient = False
 
@@ -477,11 +472,10 @@ class RecurrentOpNoMemBootTest(RecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
 
@@ -582,34 +576,30 @@ class RecurrentOpSubBlockTest(RecurrentOpTest1):
             self.output = paddle.mean(rnn_out)
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
             name='x',
-            append_batch_size=False,
         )
         x.stop_gradient = False
 
-        emb = layers.data(
+        emb = paddle.static.data(
             name='emb',
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype='float32',
-            append_batch_size=False,
         )
         emb.stop_gradient = False
 
-        w1 = layers.data(
+        w1 = paddle.static.data(
             shape=[self.input_dim, self.input_dim],
             dtype='float32',
             name='w1',
-            append_batch_size=False,
         )
         w1.stop_gradient = False
-        w2 = layers.data(
+        w2 = paddle.static.data(
             shape=[self.input_dim * 2, self.input_dim],
             dtype='float32',
             name='w2',
-            append_batch_size=False,
         )
         w2.stop_gradient = False
 
@@ -675,15 +665,14 @@ class RecurrentOpStopGradientTest(RecurrentOpTest1):
             self.output = paddle.mean(self.create_rnn_op())
 
     def create_rnn_op(self):
-        x = layers.data(
+        x = paddle.static.data(
             shape=[self.sent_len, self.batch_size, self.input_dim],
             dtype="float32",
             name="x",
-            append_batch_size=False,
         )
         x.stop_gradient = False
-        h_boot = layers.data(
-            shape=[self.input_dim], dtype="float32", name="h_boot"
+        h_boot = paddle.static.data(
+            shape=[-1, self.input_dim], dtype="float32", name="h_boot"
         )
         h_boot.stop_gradient = True
 
diff --git a/python/paddle/fluid/tests/unittests/test_reduce_op.py b/python/paddle/fluid/tests/unittests/test_reduce_op.py
index b101c24e77f8a08b33028aa7efdb586cd41a14cf..026554f010d447453a77e8c957549fc370f4d643 100644
--- a/python/paddle/fluid/tests/unittests/test_reduce_op.py
+++ b/python/paddle/fluid/tests/unittests/test_reduce_op.py
@@ -507,8 +507,8 @@ class TestAllOpError(unittest.TestCase):
             input1 = 12
             self.assertRaises(TypeError, paddle.all, input1)
             # The input dtype of reduce_all_op must be bool.
-            input2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            input2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.all, input2)
 
@@ -624,8 +624,8 @@ class TestAnyOpError(unittest.TestCase):
             input1 = 12
             self.assertRaises(TypeError, paddle.any, input1)
             # The input dtype of reduce_any_op must be bool.
-            input2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            input2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.any, input2)
 
@@ -918,7 +918,7 @@ class TestReduceSumOpError(unittest.TestCase):
             )
             self.assertRaises(TypeError, paddle.sum, x1)
             # The input dtype of reduce_sum_op  must be float32 or float64 or int32 or int64.
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="uint8")
             self.assertRaises(TypeError, paddle.sum, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_registry.py b/python/paddle/fluid/tests/unittests/test_registry.py
index 1427d0e07548aa32aa2543e8c3388cbf0126c2ba..4d40eb9965f78134f9c6683ec9ee1cc17ef4276e 100644
--- a/python/paddle/fluid/tests/unittests/test_registry.py
+++ b/python/paddle/fluid/tests/unittests/test_registry.py
@@ -24,7 +24,7 @@ import paddle.fluid as fluid
 class TestRegistry(unittest.TestCase):
     @prog_scope()
     def test_registry_layer(self):
-        x = fluid.layers.data(name='X', shape=[10, 10], dtype='float32')
+        x = paddle.static.data(name='X', shape=[-1, 10, 10], dtype='float32')
         output = paddle.mean(x)
 
         place = fluid.CPUPlace()
diff --git a/python/paddle/fluid/tests/unittests/test_regularizer.py b/python/paddle/fluid/tests/unittests/test_regularizer.py
index 5ab643819d7665ab427b5bc0bcefa69d25e2d2e5..13420553f1fe980c90f6f55a06a03dd84ed0d086 100644
--- a/python/paddle/fluid/tests/unittests/test_regularizer.py
+++ b/python/paddle/fluid/tests/unittests/test_regularizer.py
@@ -198,10 +198,12 @@ class TestRegularizer(unittest.TestCase):
         with self.scope_prog_guard(
             main_prog=main_prog, startup_prog=startup_prog
         ):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
 
             avg_cost = model(data, label, self.word_len)
 
@@ -221,10 +223,12 @@ class TestRegularizer(unittest.TestCase):
         with self.scope_prog_guard(
             main_prog=main_prog, startup_prog=startup_prog
         ):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
 
             avg_cost_l2 = model(data, label, self.word_len)
 
diff --git a/python/paddle/fluid/tests/unittests/test_regularizer_api.py b/python/paddle/fluid/tests/unittests/test_regularizer_api.py
index a863ed45fdc17f6a2f5d1e120dbcb0c745883ff1..0d3e99a6c7f1d40f05519b520c7c7d7b855960f7 100644
--- a/python/paddle/fluid/tests/unittests/test_regularizer_api.py
+++ b/python/paddle/fluid/tests/unittests/test_regularizer_api.py
@@ -105,10 +105,12 @@ class TestRegularizer(unittest.TestCase):
         with self.scope_prog_guard(
             main_prog=main_prog, startup_prog=startup_prog
         ):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
 
             avg_cost = model(data, label, self.word_len)
 
@@ -129,10 +131,12 @@ class TestRegularizer(unittest.TestCase):
         with self.scope_prog_guard(
             main_prog=main_prog, startup_prog=startup_prog
         ):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
 
             avg_cost_l2 = model(data, label, self.word_len)
 
diff --git a/python/paddle/fluid/tests/unittests/test_renorm_op.py b/python/paddle/fluid/tests/unittests/test_renorm_op.py
index 7a4cac34ecdae54d6fbfaa5942d52f1058b152e7..79dc144ee57088e35fcd539c26ea2093e02542fb 100644
--- a/python/paddle/fluid/tests/unittests/test_renorm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_renorm_op.py
@@ -38,7 +38,6 @@ class TestRenormAPI(unittest.TestCase):
 
         # case 1:
         with program_guard(Program(), Program()):
-            # x = fluid.layers.data(name = 'x',shape=[-1, 2, 3])
             x = paddle.static.data(name="x", shape=[-1, 2, 3], dtype='float64')
             z = paddle.renorm(x, self.p, self.dim, self.max_norm)
             exe = fluid.Executor(fluid.CPUPlace())
diff --git a/python/paddle/fluid/tests/unittests/test_repeat_interleave_op.py b/python/paddle/fluid/tests/unittests/test_repeat_interleave_op.py
index 6b602fa74131ef3c3e49c1ff0c7edb345dee753c..90877a3047e2c9f52ce8a58a9d93675f57156d8f 100644
--- a/python/paddle/fluid/tests/unittests/test_repeat_interleave_op.py
+++ b/python/paddle/fluid/tests/unittests/test_repeat_interleave_op.py
@@ -120,13 +120,14 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         # case 1:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
-            index = fluid.layers.data(
+            x = paddle.static.data(name='x', shape=[-1, 4], dtype='float32')
+            x.desc.set_need_check_feed(False)
+            index = paddle.static.data(
                 name='repeats_',
                 shape=[4],
                 dtype='int32',
-                append_batch_size=False,
             )
+            index.desc.set_need_check_feed(False)
             z = paddle.repeat_interleave(x, index, axis=1)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -140,13 +141,14 @@ class TestIndexSelectAPI(unittest.TestCase):
         # case 2:
         repeats = np.array([1, 2, 1]).astype('int32')
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
-            index = fluid.layers.data(
+            x = paddle.static.data(name='x', shape=[-1, 4], dtype="float32")
+            x.desc.set_need_check_feed(False)
+            index = paddle.static.data(
                 name='repeats_',
                 shape=[3],
                 dtype='int32',
-                append_batch_size=False,
             )
+            index.desc.set_need_check_feed(False)
             z = paddle.repeat_interleave(x, index, axis=0)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -162,7 +164,8 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         repeats = 2
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
+            x = paddle.static.data(name='x', shape=[-1, 4], dtype='float32')
+            x.desc.set_need_check_feed(False)
             z = paddle.repeat_interleave(x, repeats, axis=0)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -173,7 +176,8 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         # case 3 zero_dim:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[])
+            x = paddle.static.data(name='x', shape=[-1], dtype="float32")
+            x.desc.set_need_check_feed(False)
             z = paddle.repeat_interleave(x, repeats)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
diff --git a/python/paddle/fluid/tests/unittests/test_rmsprop_op.py b/python/paddle/fluid/tests/unittests/test_rmsprop_op.py
index 01a58266aebd46c02a035e5f2369ba4036ac72b2..63b22fb9fc99834ebc9aea3c44f96e2315e3cda0 100644
--- a/python/paddle/fluid/tests/unittests/test_rmsprop_op.py
+++ b/python/paddle/fluid/tests/unittests/test_rmsprop_op.py
@@ -277,8 +277,8 @@ class TestRMSPropV2(unittest.TestCase):
         place = fluid.CPUPlace()
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[1], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 1], dtype='float32')
             y_predict = paddle.static.nn.fc(x, size=1)
             cost = paddle.nn.functional.square_error_cost(
                 input=y_predict, label=y
diff --git a/python/paddle/fluid/tests/unittests/test_rnn_cell_api.py b/python/paddle/fluid/tests/unittests/test_rnn_cell_api.py
index 9c9c5520c027d8b7444e7a7343700bce23dbc1a5..730daa1cbeb4376d5b3266c9d8151c59fc725643 100644
--- a/python/paddle/fluid/tests/unittests/test_rnn_cell_api.py
+++ b/python/paddle/fluid/tests/unittests/test_rnn_cell_api.py
@@ -22,7 +22,6 @@ from rnn.rnn_numpy import rnn as numpy_rnn
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 import paddle.fluid.layers.utils as utils
 from paddle.fluid import framework
 from paddle.fluid.executor import Executor
@@ -42,10 +41,9 @@ class TestRnnError(unittest.TestCase):
             inputs = fluid.data(
                 name='inputs', shape=[None, input_size], dtype='float32'
             )
-            pre_hidden = layers.data(
+            pre_hidden = paddle.static.data(
                 name='pre_hidden',
                 shape=[None, hidden_size],
-                append_batch_size=False,
                 dtype='float32',
             )
             inputs_basic_lstm = fluid.data(
diff --git a/python/paddle/fluid/tests/unittests/test_roll_op.py b/python/paddle/fluid/tests/unittests/test_roll_op.py
index 1bb4e3392617b0b706ffb303a345e3832aefdea8..9156f728940f76b15f36915e79871a39e21635d1 100644
--- a/python/paddle/fluid/tests/unittests/test_roll_op.py
+++ b/python/paddle/fluid/tests/unittests/test_roll_op.py
@@ -68,7 +68,8 @@ class TestRollAPI(unittest.TestCase):
         paddle.enable_static()
         # case 1:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 3])
+            x = paddle.static.data(name='x', shape=[-1, 3], dtype='float32')
+            x.desc.set_need_check_feed(False)
             z = paddle.roll(x, shifts=1)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -81,7 +82,8 @@ class TestRollAPI(unittest.TestCase):
 
         # case 2:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 3])
+            x = paddle.static.data(name='x', shape=[-1, 3], dtype='float32')
+            x.desc.set_need_check_feed(False)
             z = paddle.roll(x, shifts=1, axis=0)
             exe = fluid.Executor(fluid.CPUPlace())
             (res,) = exe.run(
@@ -119,7 +121,8 @@ class TestRollAPI(unittest.TestCase):
 
         def test_axis_out_range():
             with program_guard(Program(), Program()):
-                x = fluid.layers.data(name='x', shape=[-1, 3])
+                x = paddle.static.data(name='x', shape=[-1, 3], dtype='float32')
+                x.desc.set_need_check_feed(False)
                 z = paddle.roll(x, shifts=1, axis=10)
                 exe = fluid.Executor(fluid.CPUPlace())
                 (res,) = exe.run(
diff --git a/python/paddle/fluid/tests/unittests/test_run_program_op.py b/python/paddle/fluid/tests/unittests/test_run_program_op.py
index 35686f843dec6f84ba4e698318404dfb40a99e94..fe012ded3993e2c861b2cb8d5281bda9ea9291e9 100644
--- a/python/paddle/fluid/tests/unittests/test_run_program_op.py
+++ b/python/paddle/fluid/tests/unittests/test_run_program_op.py
@@ -460,8 +460,8 @@ class TestRunProgramOpWithEmbedding(RunProgramOpTest):
 
     def build_model(self):
         # 1. simple model
-        x = fluid.layers.data(
-            name=self.input_names['X'][0], shape=[5], dtype='int64'
+        x = paddle.static.data(
+            name=self.input_names['X'][0], shape=[-1, 5], dtype='int64'
         )
         emb = fluid.input.embedding(
             input=x,
diff --git a/python/paddle/fluid/tests/unittests/test_save_model_without_var.py b/python/paddle/fluid/tests/unittests/test_save_model_without_var.py
index 1394006807029640ddbac9b003bf61ca7aa16313..97961b3df0af9bc1666098d64513256c50246b8b 100644
--- a/python/paddle/fluid/tests/unittests/test_save_model_without_var.py
+++ b/python/paddle/fluid/tests/unittests/test_save_model_without_var.py
@@ -15,14 +15,13 @@
 import unittest
 import warnings
 
+import paddle
 import paddle.fluid as fluid
 
 
 class TestSaveModelWithoutVar(unittest.TestCase):
     def test_no_var_save(self):
-        data = fluid.layers.data(
-            name='data', shape=[-1, 1], dtype='float32', append_batch_size=False
-        )
+        data = paddle.static.data(name='data', shape=[-1, 1], dtype='float32')
         data_plus = data + 1
 
         if fluid.core.is_compiled_with_cuda():
diff --git a/python/paddle/fluid/tests/unittests/test_scale_op.py b/python/paddle/fluid/tests/unittests/test_scale_op.py
index 3fc7aedf0221c2408445d0c3a89f082626b9a749..f66ff39f1087248784d2bd941e930bc5afc6bf1d 100644
--- a/python/paddle/fluid/tests/unittests/test_scale_op.py
+++ b/python/paddle/fluid/tests/unittests/test_scale_op.py
@@ -22,7 +22,6 @@ from op_test import OpTest, convert_float_to_uint16
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid.op import Operator
 from paddle.static import Program, program_guard
 
@@ -247,7 +246,7 @@ class TestScaleDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3], False, dtype)
+        data = paddle.static.data('data', [2, 3], dtype)
         data.persistable = True
         out = paddle.scale(data, 2.0)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -278,7 +277,7 @@ class TestScaleTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3], False, dtype)
+        data = paddle.static.data('data', [2, 3], dtype)
         data.persistable = True
         out = paddle.scale(data, 2.0)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_scatter_nd_op.py b/python/paddle/fluid/tests/unittests/test_scatter_nd_op.py
index b0c58839b3df22cbf267c06f606d8115aae9c63a..5c5009984ee5b992e9b85f2cb322db11f7e9f9f2 100644
--- a/python/paddle/fluid/tests/unittests/test_scatter_nd_op.py
+++ b/python/paddle/fluid/tests/unittests/test_scatter_nd_op.py
@@ -167,44 +167,38 @@ class TestScatterNdOpAPI(unittest.TestCase):
     """
 
     def testcase1(self):
-        ref1 = fluid.layers.data(
+        ref1 = paddle.static.data(
             name='ref1',
             shape=[10, 9, 8, 1, 3],
             dtype='float32',
-            append_batch_size=False,
         )
-        index1 = fluid.layers.data(
+        index1 = paddle.static.data(
             name='index1',
             shape=[5, 5, 8, 5],
             dtype='int32',
-            append_batch_size=False,
         )
-        updates1 = fluid.layers.data(
+        updates1 = paddle.static.data(
             name='update1',
             shape=[5, 5, 8],
             dtype='float32',
-            append_batch_size=False,
         )
         output1 = paddle.scatter_nd_add(ref1, index1, updates1)
 
     def testcase2(self):
-        ref2 = fluid.layers.data(
+        ref2 = paddle.static.data(
             name='ref2',
             shape=[10, 9, 8, 1, 3],
             dtype='double',
-            append_batch_size=False,
         )
-        index2 = fluid.layers.data(
+        index2 = paddle.static.data(
             name='index2',
             shape=[5, 8, 5],
             dtype='int32',
-            append_batch_size=False,
         )
-        updates2 = fluid.layers.data(
+        updates2 = paddle.static.data(
             name='update2',
             shape=[5, 8],
             dtype='double',
-            append_batch_size=False,
         )
         output2 = paddle.scatter_nd_add(
             ref2, index2, updates2, name="scatter_nd_add"
@@ -212,33 +206,29 @@ class TestScatterNdOpAPI(unittest.TestCase):
 
     def testcase3(self):
         shape3 = [10, 9, 8, 1, 3]
-        index3 = fluid.layers.data(
+        index3 = paddle.static.data(
             name='index3',
             shape=[5, 5, 8, 5],
             dtype='int32',
-            append_batch_size=False,
         )
-        updates3 = fluid.layers.data(
+        updates3 = paddle.static.data(
             name='update3',
             shape=[5, 5, 8],
             dtype='float32',
-            append_batch_size=False,
         )
         output3 = paddle.scatter_nd(index3, updates3, shape3)
 
     def testcase4(self):
         shape4 = [10, 9, 8, 1, 3]
-        index4 = fluid.layers.data(
+        index4 = paddle.static.data(
             name='index4',
             shape=[5, 5, 8, 5],
             dtype='int32',
-            append_batch_size=False,
         )
-        updates4 = fluid.layers.data(
+        updates4 = paddle.static.data(
             name='update4',
             shape=[5, 5, 8],
             dtype='double',
-            append_batch_size=False,
         )
         output4 = paddle.scatter_nd(index4, updates4, shape4, name='scatter_nd')
 
@@ -298,14 +288,14 @@ class TestScatterNdOpRaise(unittest.TestCase):
     def test_check_raise(self):
         def check_raise_is_test():
             try:
-                ref5 = fluid.layers.data(
-                    name='ref5', shape=[3, 4, 5], dtype='float32'
+                ref5 = paddle.static.data(
+                    name='ref5', shape=[-1, 3, 4, 5], dtype='float32'
                 )
-                index5 = fluid.layers.data(
-                    name='index5', shape=[2, 10], dtype='int32'
+                index5 = paddle.static.data(
+                    name='index5', shape=[-1, 2, 10], dtype='int32'
                 )
-                updates5 = fluid.layers.data(
-                    name='updates5', shape=[2, 10], dtype='float32'
+                updates5 = paddle.static.data(
+                    name='updates5', shape=[-1, 2, 10], dtype='float32'
                 )
                 output5 = paddle.scatter_nd_add(ref5, index5, updates5)
             except Exception as e:
@@ -317,23 +307,20 @@ class TestScatterNdOpRaise(unittest.TestCase):
 
     def test_check_raise2(self):
         with self.assertRaises(ValueError):
-            ref6 = fluid.layers.data(
+            ref6 = paddle.static.data(
                 name='ref6',
                 shape=[10, 9, 8, 1, 3],
                 dtype='double',
-                append_batch_size=False,
             )
-            index6 = fluid.layers.data(
+            index6 = paddle.static.data(
                 name='index6',
                 shape=[5, 8, 5],
                 dtype='int32',
-                append_batch_size=False,
             )
-            updates6 = fluid.layers.data(
+            updates6 = paddle.static.data(
                 name='update6',
                 shape=[5, 8],
                 dtype='float32',
-                append_batch_size=False,
             )
             output6 = paddle.scatter_nd_add(ref6, index6, updates6)
 
@@ -341,11 +328,11 @@ class TestScatterNdOpRaise(unittest.TestCase):
         def check_raise_is_test():
             try:
                 shape = [3, 4, 5]
-                index7 = fluid.layers.data(
-                    name='index7', shape=[2, 1], dtype='int32'
+                index7 = paddle.static.data(
+                    name='index7', shape=[-1, 2, 1], dtype='int32'
                 )
-                updates7 = fluid.layers.data(
-                    name='updates7', shape=[2, 4, 5, 20], dtype='float32'
+                updates7 = paddle.static.data(
+                    name='updates7', shape=[-1, 2, 4, 5, 20], dtype='float32'
                 )
                 output7 = paddle.scatter_nd(index7, updates7, shape)
             except Exception as e:
diff --git a/python/paddle/fluid/tests/unittests/test_select_input_output_op.py b/python/paddle/fluid/tests/unittests/test_select_input_output_op.py
index c294e5f6e50bff9aeed8180683b19123d8dcc1d1..d92688d108bd1cfb24a38c0ccd7f994fb0a3a296 100644
--- a/python/paddle/fluid/tests/unittests/test_select_input_output_op.py
+++ b/python/paddle/fluid/tests/unittests/test_select_input_output_op.py
@@ -19,7 +19,6 @@ import numpy as np
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid.backward import append_backward
 from paddle.fluid.executor import Executor
 from paddle.fluid.framework import Program, program_guard
@@ -33,9 +32,11 @@ class TestSplitMergeSelectedVarOps(unittest.TestCase):
         for branch_num in range(2, 10):
             program = Program()
             with program_guard(program):
-                x = layers.data(name='x', shape=[2], dtype='float32')
+                x = paddle.static.data(name='x', shape=[-1, 2], dtype='float32')
                 x.stop_gradient = False  # For test gradient
-                mask = layers.data(name='mask', shape=[1], dtype='int32')
+                mask = paddle.static.data(
+                    name='mask', shape=[-1, 1], dtype='int32'
+                )
 
                 outputs = []
                 for i in range(branch_num):
@@ -78,8 +79,8 @@ class TestSplitMergeSelectedVarOps(unittest.TestCase):
 class TestSelectInputOpError(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
-            mask = layers.data(name='mask', shape=[1], dtype='int32')
-            in1 = layers.data(name='in1', shape=[1], dtype='int32')
+            mask = paddle.static.data(name='mask', shape=[-1, 1], dtype='int32')
+            in1 = paddle.static.data(name='in1', shape=[-1, 1], dtype='int32')
 
             # 1. The type of inputs in select_input must be list or tuple.
             def test_inputs_type():
@@ -95,7 +96,9 @@ class TestSelectInputOpError(unittest.TestCase):
 
             # 3. The dtype of mask in select_input must be int32 or int64.
             def test_mask_dtype():
-                mask = layers.data(name='mask2', shape=[1], dtype='float32')
+                mask = paddle.static.data(
+                    name='mask2', shape=[-1, 1], dtype='float32'
+                )
                 select_input([in1], mask)
 
             self.assertRaises(TypeError, test_mask_dtype)
@@ -105,14 +108,14 @@ class TestSelectOutput_Error(unittest.TestCase):
     def test_errors(self):
         with program_guard(Program(), Program()):
 
-            in1 = layers.data(name='in1', shape=[1], dtype='int32')
-            mask_int32 = layers.data(
-                name='mask_int32', shape=[1], dtype='int32'
+            in1 = paddle.static.data(name='in1', shape=[-1, 1], dtype='int32')
+            mask_int32 = paddle.static.data(
+                name='mask_int32', shape=[-1, 1], dtype='int32'
             )
-            mask_float32 = layers.data(
-                name='mask_float32', shape=[1], dtype='float32'
+            mask_float32 = paddle.static.data(
+                name='mask_float32', shape=[-1, 1], dtype='float32'
             )
-            out1 = layers.data(name='out1', shape=[1], dtype='int32')
+            out1 = paddle.static.data(name='out1', shape=[-1, 1], dtype='int32')
 
             # 1. The type of input in select_output must Variable.
             def test_input_type():
diff --git a/python/paddle/fluid/tests/unittests/test_set_bool_attr.py b/python/paddle/fluid/tests/unittests/test_set_bool_attr.py
index da02e4621d0c23383b56e93899117e86eeaf7647..c599f08ae2bf99b7747a007ec0fdb6aadfc81b0a 100644
--- a/python/paddle/fluid/tests/unittests/test_set_bool_attr.py
+++ b/python/paddle/fluid/tests/unittests/test_set_bool_attr.py
@@ -21,7 +21,9 @@ import paddle.fluid as fluid
 
 class TestAttrSet(unittest.TestCase):
     def test_set_bool_attr(self):
-        x = fluid.layers.data(name='x', shape=[3, 7, 3, 7], dtype='float32')
+        x = paddle.static.data(
+            name='x', shape=[-1, 3, 7, 3, 7], dtype='float32'
+        )
         param_attr = fluid.ParamAttr(
             name='batch_norm_w',
             initializer=fluid.initializer.Constant(value=1.0),
diff --git a/python/paddle/fluid/tests/unittests/test_sgd_op_bf16.py b/python/paddle/fluid/tests/unittests/test_sgd_op_bf16.py
index 81b8c1b2b164374ffff6a62e438909ecac53c8f8..89515c931c2504c8fc2d915308f3dc9f3b069bdc 100644
--- a/python/paddle/fluid/tests/unittests/test_sgd_op_bf16.py
+++ b/python/paddle/fluid/tests/unittests/test_sgd_op_bf16.py
@@ -334,9 +334,13 @@ class TestSGDOpBF16API(unittest.TestCase):
         place = fluid.CPUPlace()
         main = fluid.Program()
         with fluid.program_guard(main):
-            x = fluid.layers.data(name='X', shape=self.ids_shape, dtype='int64')
-            label = fluid.layers.data(
-                name='Y', shape=self.y_shape, dtype='uint16'
+            ids_shape = list(self.ids_shape)
+            x = paddle.static.data(
+                name='X', shape=[-1] + ids_shape, dtype='int64'
+            )
+            y_shape = list(self.y_shape)
+            label = paddle.static.data(
+                name='Y', shape=[-1] + y_shape, dtype='uint16'
             )
             emb = fluid.layers.embedding(
                 input=x,
diff --git a/python/paddle/fluid/tests/unittests/test_sigmoid_cross_entropy_with_logits_op.py b/python/paddle/fluid/tests/unittests/test_sigmoid_cross_entropy_with_logits_op.py
index ce18a4647966d37eb7953742d48581d8d1afe2f6..de6e6bd725aea6c94b70f192be1a93dffb76a749 100644
--- a/python/paddle/fluid/tests/unittests/test_sigmoid_cross_entropy_with_logits_op.py
+++ b/python/paddle/fluid/tests/unittests/test_sigmoid_cross_entropy_with_logits_op.py
@@ -286,11 +286,11 @@ class TestSigmoidCrossEntropyWithNorm2(OpTest):
                 def test_dtype():
                     # the input dtype of sigmoid_cross_entropy_with_logits must be float16 or float32 or float64
                     # float16 only can be set on GPU place
-                    x2 = fluid.layers.data(
-                        name='x2', shape=[3, 4, 5, 6], dtype="int32"
+                    x2 = paddle.static.data(
+                        name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                     )
-                    lab2 = fluid.layers.data(
-                        name='lab2', shape=[3, 4, 5, 6], dtype="int32"
+                    lab2 = paddle.static.data(
+                        name='lab2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                     )
                     paddle.nn.functional.binary_cross_entropy_with_logits(
                         x2, lab2
diff --git a/python/paddle/fluid/tests/unittests/test_sign_op.py b/python/paddle/fluid/tests/unittests/test_sign_op.py
index ca6ca55c91f9713c56bc76bb4ca20a292e278bf6..7834736260d9dabd7b769f29f8416bcb6384b43e 100644
--- a/python/paddle/fluid/tests/unittests/test_sign_op.py
+++ b/python/paddle/fluid/tests/unittests/test_sign_op.py
@@ -22,7 +22,6 @@ from op_test import OpTest
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 
 
@@ -48,16 +47,16 @@ class TestSignOpError(unittest.TestCase):
             input1 = 12
             self.assertRaises(TypeError, paddle.sign, input1)
             # The input dtype of sign_op must be float16, float32, float64.
-            input2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            input2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
-            input3 = fluid.layers.data(
-                name='input3', shape=[12, 10], dtype="int64"
+            input3 = paddle.static.data(
+                name='input3', shape=[-1, 12, 10], dtype="int64"
             )
             self.assertRaises(TypeError, paddle.sign, input2)
             self.assertRaises(TypeError, paddle.sign, input3)
-            input4 = fluid.layers.data(
-                name='input4', shape=[4], dtype="float16"
+            input4 = paddle.static.data(
+                name='input4', shape=[-1, 4], dtype="float16"
             )
             paddle.sign(input4)
 
@@ -78,16 +77,16 @@ class TestSignAPI(unittest.TestCase):
             input1 = 12
             self.assertRaises(TypeError, paddle.tensor.math.sign, input1)
             # The input dtype of sign_op must be float16, float32, float64.
-            input2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            input2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
-            input3 = fluid.layers.data(
-                name='input3', shape=[12, 10], dtype="int64"
+            input3 = paddle.static.data(
+                name='input3', shape=[-1, 12, 10], dtype="int64"
             )
             self.assertRaises(TypeError, paddle.tensor.math.sign, input2)
             self.assertRaises(TypeError, paddle.tensor.math.sign, input3)
-            input4 = fluid.layers.data(
-                name='input4', shape=[4], dtype="float16"
+            input4 = paddle.static.data(
+                name='input4', shape=[-1, 4], dtype="float16"
             )
             paddle.sign(input4)
 
@@ -102,7 +101,7 @@ class TestSignDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [1, 4], False, dtype)
+        data = paddle.static.data('data', [1, 4], dtype)
         data.persistable = True
         out = paddle.sign(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -133,7 +132,7 @@ class TestSignTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [1, 4], False, dtype)
+        data = paddle.static.data('data', [1, 4], dtype)
         data.persistable = True
         out = paddle.sign(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_slice_op.py b/python/paddle/fluid/tests/unittests/test_slice_op.py
index 12838b218b43eadb6c2d8e45cfde4c7094a33a2d..19aa669badf5c48bfc0720c1a053fd6f5bd50bde 100644
--- a/python/paddle/fluid/tests/unittests/test_slice_op.py
+++ b/python/paddle/fluid/tests/unittests/test_slice_op.py
@@ -22,7 +22,6 @@ from op_test import OpTest, convert_float_to_uint16
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.tensor.manipulation import tensor_array_to_tensor
 
 paddle.enable_static()
@@ -551,17 +550,15 @@ class TestSliceAPI(unittest.TestCase):
         input = np.random.random([3, 4, 5, 6]).astype("float64")
         minus_1 = fluid.layers.fill_constant([1], "int32", -1)
         minus_3 = fluid.layers.fill_constant([1], "int64", -3)
-        starts = fluid.layers.data(
-            name='starts', shape=[1, 3], append_batch_size=False
+        starts = paddle.static.data(
+            name='starts', shape=[1, 3], dtype="float32"
         )
-        ends = fluid.layers.data(
-            name='ends', shape=[3], append_batch_size=False
-        )
-
-        x = fluid.layers.data(
+        starts.desc.set_need_check_feed(False)
+        ends = paddle.static.data(name='ends', shape=[3], dtype="float32")
+        ends.desc.set_need_check_feed(False)
+        x = paddle.static.data(
             name="x",
             shape=[3, 4, 5, 6],
-            append_batch_size=False,
             dtype="float64",
         )
 
@@ -885,7 +882,7 @@ class TestSliceDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [4, 5, 6], False, dtype)
+        data = paddle.static.data('data', [4, 5, 6], dtype)
         data.persistable = True
         out = paddle.slice(
             data, axes=[0, 1, 2], starts=[-3, 0, 2], ends=[3, 2, 4]
@@ -920,7 +917,7 @@ class TestSliceTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [4, 5, 6], False, dtype)
+        data = paddle.static.data('data', [4, 5, 6], dtype)
         data.persistable = True
         out = paddle.slice(
             data, axes=[0, 1, 2], starts=[-3, 0, 2], ends=[3, 2, 4]
diff --git a/python/paddle/fluid/tests/unittests/test_split_op.py b/python/paddle/fluid/tests/unittests/test_split_op.py
index 943861ab2f57b3c06d3eaf18a5f530d08648109e..40e7bff55e0bcc85f97c4f658bf91433ed083e07 100644
--- a/python/paddle/fluid/tests/unittests/test_split_op.py
+++ b/python/paddle/fluid/tests/unittests/test_split_op.py
@@ -304,34 +304,46 @@ class TestSplitOpError(unittest.TestCase):
         with program_guard(Program(), Program()):
             # The type of axis in split_op should be int or Variable.
             def test_axis_type():
-                x6 = fluid.layers.data(shape=[4], dtype='float16', name='x3')
+                x6 = paddle.static.data(
+                    shape=[-1, 4], dtype='float16', name='x3'
+                )
                 paddle.split(x=x6, num_or_sections=2, axis=3.2)
 
             self.assertRaises(TypeError, test_axis_type)
 
             # The type of axis in split_op should be int or Variable.
             def test_axis_variable_type():
-                x9 = fluid.layers.data(shape=[4], dtype='float16', name='x9')
-                x10 = fluid.layers.data(shape=[1], dtype='float16', name='x10')
+                x9 = paddle.static.data(
+                    shape=[-1, 4], dtype='float16', name='x9'
+                )
+                x10 = paddle.static.data(
+                    shape=[-1, 1], dtype='float16', name='x10'
+                )
                 paddle.split(x=x9, num_or_sections=2, axis=x10)
 
             self.assertRaises(TypeError, test_axis_variable_type)
 
             # The type of num_or_sections in split_op should be int, tuple or list.
             def test_num_or_sections_type():
-                x6 = fluid.layers.data(shape=[4], dtype='float16', name='x4')
+                x6 = paddle.static.data(
+                    shape=[-1, 4], dtype='float16', name='x4'
+                )
                 paddle.split(x=x6, num_or_sections=2.1, axis=3)
 
             self.assertRaises(TypeError, test_num_or_sections_type)
 
             def test_num_or_sections_type_tensor():
-                x7 = fluid.layers.data(shape=[4], dtype='float16', name='x5')
+                x7 = paddle.static.data(
+                    shape=[-1, 4], dtype='float16', name='x5'
+                )
                 paddle.split(input=x7, num_or_sections=2.1, dim=3)
 
             self.assertRaises(TypeError, test_num_or_sections_type_tensor)
 
             def test_axis_type_tensor():
-                x8 = fluid.layers.data(shape=[4], dtype='float16', name='x6')
+                x8 = paddle.static.data(
+                    shape=[-1, 4], dtype='float16', name='x6'
+                )
                 paddle.split(input=x8, num_or_sections=2, dim=3.2)
 
             self.assertRaises(TypeError, test_axis_type_tensor)
@@ -340,8 +352,12 @@ class TestSplitOpError(unittest.TestCase):
 class API_TestSplit(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[4, 6, 6], dtype='float64')
-            data2 = fluid.layers.data('data2', shape=[1], dtype='int32')
+            data1 = paddle.static.data(
+                'data1', shape=[-1, 4, 6, 6], dtype='float64'
+            )
+            data1.desc.set_need_check_feed(False)
+            data2 = paddle.static.data('data2', shape=[-1, 1], dtype='int32')
+            data2.desc.set_need_check_feed(False)
             x0, x1, x2 = paddle.split(data1, num_or_sections=3, axis=data2)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
@@ -359,7 +375,10 @@ class API_TestSplit(unittest.TestCase):
 class API_TestSplit2(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[4, 6, 6], dtype='float64')
+            data1 = paddle.static.data(
+                'data1', shape=[-1, 4, 6, 6], dtype='float64'
+            )
+            data1.desc.set_need_check_feed(False)
             x0, x1, x2 = paddle.split(data1, num_or_sections=3, axis=2)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
@@ -378,7 +397,7 @@ class API_TestSplit2(unittest.TestCase):
 class API_TestSplit3(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float64')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float64')
             x0, x1 = paddle.split(data, num_or_sections=(3, 7), axis=1)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
@@ -392,8 +411,8 @@ class API_TestSplit3(unittest.TestCase):
 class API_TestSplit4(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float64')
-            index = fluid.layers.data('index', shape=[1], dtype='int32')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float64')
+            index = paddle.static.data('index', shape=[1], dtype='int32')
             x0, x1 = paddle.split(data, num_or_sections=(3, index), axis=1)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
@@ -430,7 +449,7 @@ class API_TestSplit5(unittest.TestCase):
 class API_TestSplit6(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data = fluid.layers.data('data', shape=[-1, 10], dtype='float64')
+            data = paddle.static.data('data', shape=[-1, 10], dtype='float64')
             x0, x1 = paddle.split(data, num_or_sections=[1, 1], axis=0)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/test_squeeze_op.py b/python/paddle/fluid/tests/unittests/test_squeeze_op.py
index b60152a51440a3e5bbc8628262a26298fb815a14..ae3b67a2f19d1390fe449b2a125763c24ef72587 100755
--- a/python/paddle/fluid/tests/unittests/test_squeeze_op.py
+++ b/python/paddle/fluid/tests/unittests/test_squeeze_op.py
@@ -22,7 +22,6 @@ from op_test import OpTest, convert_float_to_uint16
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 
 paddle.enable_static()
@@ -228,7 +227,7 @@ class TestSqueezeDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3], False, dtype)
+        data = paddle.static.data('data', [2, 3], dtype)
         data.persistable = True
         out = paddle.squeeze(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -259,7 +258,7 @@ class TestSqueezeTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3], False, dtype)
+        data = paddle.static.data('data', [2, 3], dtype)
         data.persistable = True
         out = paddle.squeeze(data)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_stack_op.py b/python/paddle/fluid/tests/unittests/test_stack_op.py
index 9e84268a5fe4850a6fcb1ca56050c16f8e2e63d9..496877dc55c367a2be14b3eddfd5050e8dcd8e28 100644
--- a/python/paddle/fluid/tests/unittests/test_stack_op.py
+++ b/python/paddle/fluid/tests/unittests/test_stack_op.py
@@ -226,9 +226,9 @@ class TestTensorStackAPIWithLoDTensorArray(unittest.TestCase):
 class API_test(unittest.TestCase):
     def test_out(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
-            data1 = fluid.layers.data('data1', shape=[1, 2], dtype='float64')
-            data2 = fluid.layers.data('data2', shape=[1, 2], dtype='float64')
-            data3 = fluid.layers.data('data3', shape=[1, 2], dtype='float64')
+            data1 = paddle.static.data('data1', shape=[1, 2], dtype='float64')
+            data2 = paddle.static.data('data2', shape=[1, 2], dtype='float64')
+            data3 = paddle.static.data('data3', shape=[1, 2], dtype='float64')
             result_stack = paddle.stack([data1, data2, data3], axis=0)
             place = fluid.CPUPlace()
             exe = fluid.Executor(place)
diff --git a/python/paddle/fluid/tests/unittests/test_static_save_load.py b/python/paddle/fluid/tests/unittests/test_static_save_load.py
index e3309e18a4870c4cda40c76e4b2cf2a51569b0b2..d043e3785c4984fd3a1fdfd70a40109dc8bb58ca 100644
--- a/python/paddle/fluid/tests/unittests/test_static_save_load.py
+++ b/python/paddle/fluid/tests/unittests/test_static_save_load.py
@@ -281,16 +281,20 @@ class TestSaveLoadBase(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -406,16 +410,20 @@ class TestSaveLoadPartial(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -544,16 +552,20 @@ class TestSaveLoadSetStateDict(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -665,16 +677,20 @@ class TestProgramStatePartial(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -992,16 +1008,20 @@ class TestLoadFromOldInterface(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -1131,17 +1151,20 @@ class TestLoadFromOldInterface(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
-
+            init_cell.desc.set_need_check_feed(False)
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
             )
@@ -1271,16 +1294,20 @@ class TestLoadFromOldInterfaceSingleFile(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -1462,16 +1489,20 @@ class TestProgramStateOldSave(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -1634,16 +1665,20 @@ class TestProgramStateOldSaveSingleModel(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = Adam(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
@@ -1769,6 +1804,7 @@ class TestStaticSaveLoadPickle(unittest.TestCase):
                 shape=[None, 10],
                 dtype='float32',
             )
+            x.desc.set_need_check_feed(False)
             z = paddle.static.nn.fc(x, 10, bias_attr=False)
             place = paddle.CPUPlace()
             exe = paddle.static.Executor(place)
@@ -1838,6 +1874,7 @@ class TestSaveLoadInferenceModel(unittest.TestCase):
         main_program = framework.Program()
         with framework.program_guard(main_program):
             x = paddle.static.data(name="x", shape=[10, 10], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = x + x
 
             place = paddle.CPUPlace()
diff --git a/python/paddle/fluid/tests/unittests/test_static_save_load_bf16.py b/python/paddle/fluid/tests/unittests/test_static_save_load_bf16.py
index 5a13133bf79f2a480e86762b1e2dfa11a1037690..024f31fdf92cc80c434786d8826108010834212c 100644
--- a/python/paddle/fluid/tests/unittests/test_static_save_load_bf16.py
+++ b/python/paddle/fluid/tests/unittests/test_static_save_load_bf16.py
@@ -65,16 +65,20 @@ class TestSaveLoadBF16(unittest.TestCase):
             place = self.set_place()
             exe = fluid.Executor(place)
             sgd = SGDOptimizer(learning_rate=1e-3)
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name="x", shape=[-1, num_steps], dtype='int64'
             )
-            y = fluid.layers.data(name="y", shape=[-1, 1], dtype='float32')
-            init_hidden = fluid.layers.data(
-                name="init_hidden", shape=[1], dtype='float32'
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name="y", shape=[-1, 1], dtype='float32')
+            y.desc.set_need_check_feed(False)
+            init_hidden = paddle.static.data(
+                name="init_hidden", shape=[-1, 1], dtype='float32'
             )
-            init_cell = fluid.layers.data(
-                name="init_cell", shape=[1], dtype='float32'
+            init_hidden.desc.set_need_check_feed(False)
+            init_cell = paddle.static.data(
+                name="init_cell", shape=[-1, 1], dtype='float32'
             )
+            init_cell.desc.set_need_check_feed(False)
 
             static_loss, static_last_hidden, static_last_cell = ptb_model(
                 x, y, init_hidden, init_cell
diff --git a/python/paddle/fluid/tests/unittests/test_static_shape_inferrence_for_shape_tensor.py b/python/paddle/fluid/tests/unittests/test_static_shape_inferrence_for_shape_tensor.py
index 5ba3bcbbc110866da51f61f5b2a6ccd3398a936d..9e2391233ce60163e664207b16441d3943978c0c 100644
--- a/python/paddle/fluid/tests/unittests/test_static_shape_inferrence_for_shape_tensor.py
+++ b/python/paddle/fluid/tests/unittests/test_static_shape_inferrence_for_shape_tensor.py
@@ -21,9 +21,7 @@ from paddle.fluid.layers.utils import try_set_static_shape_tensor
 class StaticShapeInferrenceTest(unittest.TestCase):
     def test_static_graph(self):
         paddle.enable_static()
-        data = paddle.fluid.layers.data(
-            name="x", shape=[-1, 2], dtype='float32'
-        )
+        data = paddle.static.data(name="x", shape=[-1, 2], dtype='float32')
         shape = paddle.shape(data)  # shape should be [-1, 2]
         x = paddle.uniform(shape)
         try_set_static_shape_tensor(x, shape)
diff --git a/python/paddle/fluid/tests/unittests/test_strided_slice_op.py b/python/paddle/fluid/tests/unittests/test_strided_slice_op.py
index 996b0c4a33338f8879c81f67af05cd38ec6d64e6..05a2631d775d0f4767d3be2d857b6a25c627dc1a 100644
--- a/python/paddle/fluid/tests/unittests/test_strided_slice_op.py
+++ b/python/paddle/fluid/tests/unittests/test_strided_slice_op.py
@@ -550,20 +550,13 @@ class TestStridedSliceAPI(unittest.TestCase):
         input = np.random.random([3, 4, 5, 6]).astype("float64")
         minus_1 = fluid.layers.fill_constant([1], "int32", -1)
         minus_3 = fluid.layers.fill_constant([1], "int32", -3)
-        starts = fluid.layers.data(
-            name='starts', shape=[3], dtype='int32', append_batch_size=False
-        )
-        ends = fluid.layers.data(
-            name='ends', shape=[3], dtype='int32', append_batch_size=False
-        )
-        strides = fluid.layers.data(
-            name='strides', shape=[3], dtype='int32', append_batch_size=False
-        )
+        starts = paddle.static.data(name='starts', shape=[3], dtype='int32')
+        ends = paddle.static.data(name='ends', shape=[3], dtype='int32')
+        strides = paddle.static.data(name='strides', shape=[3], dtype='int32')
 
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[3, 4, 5, 6],
-            append_batch_size=False,
             dtype="float64",
         )
         out_1 = paddle.strided_slice(
diff --git a/python/paddle/fluid/tests/unittests/test_sum_op.py b/python/paddle/fluid/tests/unittests/test_sum_op.py
index 22b9dc573fb93aa97df5c1aed8ea6a13b9d7b2c6..6e9ff86cb8b7f83ad52c747375dd285ae989b2bc 100644
--- a/python/paddle/fluid/tests/unittests/test_sum_op.py
+++ b/python/paddle/fluid/tests/unittests/test_sum_op.py
@@ -23,7 +23,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 import paddle.inference as paddle_infer
 from paddle import enable_static
 from paddle.fluid.op import Operator
@@ -603,9 +602,9 @@ class TestAddNDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data1 = layers.data('data1', [3, 4, 5], False, dtype)
+        data1 = paddle.static.data('data1', [3, 4, 5], dtype)
         data1.persistable = True
-        data2 = layers.data('data2', [3, 4, 5], False, dtype)
+        data2 = paddle.static.data('data2', [3, 4, 5], dtype)
         data2.persistable = True
         out = paddle.add_n([data1, data2])
         data1_arr = np.random.uniform(-1, 1, data1.shape).astype(dtype)
@@ -645,9 +644,9 @@ class TestAddNTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data1 = layers.data('data1', [3, 4, 5], False, dtype)
+        data1 = paddle.static.data('data1', [3, 4, 5], dtype)
         data1.persistable = True
-        data2 = layers.data('data2', [3, 4, 5], False, dtype)
+        data2 = paddle.static.data('data2', [3, 4, 5], dtype)
         data2.persistable = True
         out = paddle.add_n([data1, data2])
         data1_arr = np.random.uniform(-1, 1, data1.shape).astype(dtype)
@@ -687,7 +686,7 @@ class TestSumDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 4], False, dtype)
+        data = paddle.static.data('data', [2, 4], dtype)
         data.persistable = True
         out = paddle.sum(data, axis=1, keepdim=True)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -718,7 +717,7 @@ class TestSumTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 4], False, dtype)
+        data = paddle.static.data('data', [2, 4], dtype)
         data.persistable = True
         out = paddle.sum(data, axis=1, keepdim=True)
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_sync_batch_norm_op.py b/python/paddle/fluid/tests/unittests/test_sync_batch_norm_op.py
index 1b8f912bb508b1658f2c01a68c1e9d3a7a72afa8..14cebe7e6461f56e23719aca0ca0e2964a0480c1 100644
--- a/python/paddle/fluid/tests/unittests/test_sync_batch_norm_op.py
+++ b/python/paddle/fluid/tests/unittests/test_sync_batch_norm_op.py
@@ -67,12 +67,12 @@ class TestSyncBatchNormOpTraining(unittest.TestCase):
         use_cudnn = self.dtype == np.float16
         with fluid.unique_name.guard():
             with fluid.program_guard(main, startup):
-                data = fluid.layers.data(
+                data = paddle.static.data(
                     name='input',
                     shape=self.dshape,
                     dtype=self.dtype,
-                    append_batch_size=False,
                 )
+                data.desc.set_need_check_feed(False)
                 conv = paddle.static.nn.conv2d(
                     input=data,
                     num_filters=32,
@@ -248,7 +248,10 @@ class TestDygraphSyncBatchNormAPIError(unittest.TestCase):
 
             # the input dtype of SyncBatchNorm must be float16 or float32 or float64
             # float16 only can be set on GPU place
-            x2 = fluid.layers.data(name='x2', shape=[3, 4, 5, 6], dtype="int32")
+            x2 = paddle.static.data(
+                name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
+            )
+            x2.desc.set_need_check_feed(False)
             self.assertRaises(TypeError, my_sync_batch_norm, x2)
 
 
diff --git a/python/paddle/fluid/tests/unittests/test_tdm_child_op.py b/python/paddle/fluid/tests/unittests/test_tdm_child_op.py
index 5d261dd1efb0b2e335fe1f559395a6ffdfd569fe..2481a48f01793cdeb8ef30b528589447fe76f6fd 100644
--- a/python/paddle/fluid/tests/unittests/test_tdm_child_op.py
+++ b/python/paddle/fluid/tests/unittests/test_tdm_child_op.py
@@ -17,6 +17,7 @@ import unittest
 import numpy as np
 from op_test import OpTest
 
+import paddle
 import paddle.fluid as fluid
 
 
@@ -139,7 +140,9 @@ class TestCase4(TestTDMChildOp):
 
 class TestTDMChildShape(unittest.TestCase):
     def test_shape(self):
-        x = fluid.layers.data(name='x', shape=[1], dtype='int32', lod_level=1)
+        x = paddle.static.data(
+            name='x', shape=[-1, 1], dtype='int32', lod_level=1
+        )
         tdm_tree_info = create_tdm_tree()
         tree_info_np = np.array(tdm_tree_info).astype('int32')
 
diff --git a/python/paddle/fluid/tests/unittests/test_tdm_sampler_op.py b/python/paddle/fluid/tests/unittests/test_tdm_sampler_op.py
index 7a649a8c238dbea509a36f40044af3ebcab6792d..217d84b4b9f8afc2d9a52adec74f50ab3e07e6d5 100644
--- a/python/paddle/fluid/tests/unittests/test_tdm_sampler_op.py
+++ b/python/paddle/fluid/tests/unittests/test_tdm_sampler_op.py
@@ -18,6 +18,7 @@ import unittest
 import numpy as np
 from op_test import OpTest
 
+import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
 
@@ -266,7 +267,9 @@ class TestCase7(TestTDMSamplerOp):
 
 class TestTDMSamplerShape(unittest.TestCase):
     def test_shape(self):
-        x = fluid.layers.data(name='x', shape=[1], dtype='int32', lod_level=1)
+        x = paddle.static.data(
+            name='x', shape=[-1, 1], dtype='int32', lod_level=1
+        )
         tdm_tree_travel = create_tdm_travel()
         tdm_tree_layer = create_tdm_layer()
         layer_node_num_list = [len(i) for i in tdm_tree_layer]
diff --git a/python/paddle/fluid/tests/unittests/test_tile_op.py b/python/paddle/fluid/tests/unittests/test_tile_op.py
index 419c142b6df5ff54909d62e649471741b48666bd..bc549917e8ba8d4f14769777e0ec82878487570d 100644
--- a/python/paddle/fluid/tests/unittests/test_tile_op.py
+++ b/python/paddle/fluid/tests/unittests/test_tile_op.py
@@ -21,7 +21,6 @@ from op_test import OpTest
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, core, program_guard
 
 
@@ -229,9 +228,9 @@ class TestTileError(unittest.TestCase):
             )
             repeat_times = [2, 2]
             self.assertRaises(TypeError, paddle.tile, x1, repeat_times)
-            x2 = fluid.layers.data(name='x2', shape=[4], dtype="uint8")
+            x2 = paddle.static.data(name='x2', shape=[-1, 4], dtype="uint8")
             self.assertRaises(TypeError, paddle.tile, x2, repeat_times)
-            x3 = fluid.layers.data(name='x3', shape=[4], dtype="bool")
+            x3 = paddle.static.data(name='x3', shape=[-1, 4], dtype="bool")
             x3.stop_gradient = False
             self.assertRaises(ValueError, paddle.tile, x3, repeat_times)
 
@@ -240,7 +239,7 @@ class TestTileAPIStatic(unittest.TestCase):
     def test_api(self):
         with program_guard(Program(), Program()):
             repeat_times = [2, 2]
-            x1 = fluid.layers.data(name='x1', shape=[4], dtype="int32")
+            x1 = paddle.static.data(name='x1', shape=[-1, 4], dtype="int32")
             out = paddle.tile(x1, repeat_times)
             positive_2 = fluid.layers.fill_constant([1], dtype="int32", value=2)
             out2 = paddle.tile(x1, repeat_times=[positive_2, 2])
@@ -278,7 +277,7 @@ class TestTileDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [1, 2], False, dtype)
+        data = paddle.static.data('data', [1, 2], dtype)
         data.persistable = True
         out = paddle.tile(data, [2, 1])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -309,7 +308,7 @@ class TestTileTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [1, 2], False, dtype)
+        data = paddle.static.data('data', [1, 2], dtype)
         data.persistable = True
         out = paddle.tile(data, [2, 1])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_trainable.py b/python/paddle/fluid/tests/unittests/test_trainable.py
index a9f96230ffb0fa050be7026671335226336e935d..86664d4f3b69508a35c3cb20abf59cd138c70f1d 100644
--- a/python/paddle/fluid/tests/unittests/test_trainable.py
+++ b/python/paddle/fluid/tests/unittests/test_trainable.py
@@ -22,8 +22,8 @@ import paddle.fluid as fluid
 
 
 def test_trainable():
-    x = fluid.layers.data(name='image', shape=[784], dtype='float32')
-    label = fluid.layers.data(name='label', shape=[1], dtype='int64')
+    x = paddle.static.data(name='image', shape=[-1, 784], dtype='float32')
+    label = paddle.static.data(name='label', shape=[-1, 1], dtype='int64')
     feature = paddle.static.nn.fc(
         x, size=10, weight_attr=fluid.ParamAttr(trainable=False)
     )
diff --git a/python/paddle/fluid/tests/unittests/test_transpose_op.py b/python/paddle/fluid/tests/unittests/test_transpose_op.py
index ad57289ffd3ebbbbceb13fea7f3099e1e26a57c9..a2f922dcd8db427b6c70666ca1b724fb9aa858a5 100644
--- a/python/paddle/fluid/tests/unittests/test_transpose_op.py
+++ b/python/paddle/fluid/tests/unittests/test_transpose_op.py
@@ -21,7 +21,6 @@ from decorator_helper import prog_scope
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 from paddle.fluid import Program, program_guard
 from paddle.fluid.tests.unittests.op_test import OpTest, convert_float_to_uint16
 
@@ -289,7 +288,9 @@ class TestTransposeOpError(unittest.TestCase):
     def test_errors(self):
         paddle.enable_static()
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[10, 5, 3], dtype='float64')
+            x = paddle.static.data(
+                name='x', shape=[-1, 10, 5, 3], dtype='float64'
+            )
 
             def test_x_Variable_check():
                 # the Input(x)'s type must be Variable
@@ -299,8 +300,8 @@ class TestTransposeOpError(unittest.TestCase):
 
             def test_x_dtype_check():
                 # the Input(x)'s dtype must be one of [bool, float16, float32, float64, int32, int64]
-                x1 = fluid.layers.data(
-                    name='x1', shape=[10, 5, 3], dtype='int8'
+                x1 = paddle.static.data(
+                    name='x1', shape=[-1, 10, 5, 3], dtype='int8'
                 )
                 paddle.transpose(x1, perm=[1, 0, 2])
 
@@ -520,7 +521,7 @@ class TestTransposeDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3, 4], False, dtype)
+        data = paddle.static.data('data', [2, 3, 4], dtype)
         data.persistable = True
         out = paddle.transpose(data, [1, 0, 2])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -551,7 +552,7 @@ class TestTransposeTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3, 4], False, dtype)
+        data = paddle.static.data('data', [2, 3, 4], dtype)
         data.persistable = True
         out = paddle.transpose(data, [1, 0, 2])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_uniform_random_op.py b/python/paddle/fluid/tests/unittests/test_uniform_random_op.py
index e86dec9899af2ebe66f7bdc9a0a2c4e34aa42031..c31d763dbff7c50d085d8d73bc93ab7ec8e4e089 100644
--- a/python/paddle/fluid/tests/unittests/test_uniform_random_op.py
+++ b/python/paddle/fluid/tests/unittests/test_uniform_random_op.py
@@ -280,7 +280,9 @@ class TestUniformRandomOpSelectedRowsWithDiagInit(
 class TestUniformRandomOpApi(unittest.TestCase):
     def test_api(self):
         paddle.seed(10)
-        x = fluid.layers.data('x', shape=[16], dtype='float32', lod_level=1)
+        x = paddle.static.data(
+            'x', shape=[-1, 16], dtype='float32', lod_level=1
+        )
         y = paddle.static.nn.fc(
             x,
             size=16,
@@ -467,16 +469,16 @@ class TestUniformRandomBatchSizeLikeOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_Variable)
 
             def test_shape():
-                x1 = fluid.layers.data(
-                    name='x2', shape=[100, 784], dtype='float32'
+                x1 = paddle.static.data(
+                    name='x2', shape=[-1, 100, 784], dtype='float32'
                 )
                 random.uniform_random_batch_size_like(x1, shape="shape")
 
             self.assertRaises(TypeError, test_shape)
 
             def test_dtype():
-                x2 = fluid.layers.data(
-                    name='x2', shape=[100, 784], dtype='float32'
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 100, 784], dtype='float32'
                 )
                 random.uniform_random_batch_size_like(x2, 'int32')
 
@@ -516,8 +518,8 @@ class TestUniformOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_Variable2)
 
             def test_dtype():
-                x2 = fluid.layers.data(
-                    name='x2', shape=[100, 784], dtype='float32'
+                x2 = paddle.static.data(
+                    name='x2', shape=[-1, 100, 784], dtype='float32'
                 )
                 paddle.tensor.random.uniform(x2, 'int32')
 
diff --git a/python/paddle/fluid/tests/unittests/test_unsqueeze_op.py b/python/paddle/fluid/tests/unittests/test_unsqueeze_op.py
index 44fd0888f6b63298be9f7d11e6beab7f79c47e78..fdb68a27795c2064c40793c0925d00c2315aa3bb 100755
--- a/python/paddle/fluid/tests/unittests/test_unsqueeze_op.py
+++ b/python/paddle/fluid/tests/unittests/test_unsqueeze_op.py
@@ -22,7 +22,6 @@ from op_test import OpTest, convert_float_to_uint16
 import paddle
 import paddle.fluid as fluid
 import paddle.fluid.core as core
-import paddle.fluid.layers as layers
 
 paddle.enable_static()
 
@@ -329,7 +328,7 @@ class TestUnsqueezeDoubleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3, 4], False, dtype)
+        data = paddle.static.data('data', [2, 3, 4], dtype)
         data.persistable = True
         out = paddle.unsqueeze(data, [0, 2])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
@@ -360,7 +359,7 @@ class TestUnsqueezeTripleGradCheck(unittest.TestCase):
         eps = 0.005
         dtype = np.float32
 
-        data = layers.data('data', [2, 3, 4], False, dtype)
+        data = paddle.static.data('data', [2, 3, 4], dtype)
         data.persistable = True
         out = paddle.unsqueeze(data, [0, 2])
         data_arr = np.random.uniform(-1, 1, data.shape).astype(dtype)
diff --git a/python/paddle/fluid/tests/unittests/test_variable.py b/python/paddle/fluid/tests/unittests/test_variable.py
index 8520cf9067c05cce8a1f943f34d418e9b2ff6aae..94c098967e44597856f077dacd47561498eb2e13 100644
--- a/python/paddle/fluid/tests/unittests/test_variable.py
+++ b/python/paddle/fluid/tests/unittests/test_variable.py
@@ -168,7 +168,7 @@ class TestVariable(unittest.TestCase):
             var14 = var[1:-1, 0:2, ::-1]
             var15 = var[::-1, ::-1, ::-1]
 
-            x = fluid.layers.data(name='x', shape=[13], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 13], dtype='float32')
             y = paddle.static.nn.fc(x, size=1, activation=None)
             y_1 = y[:, 0]
             feeder = fluid.DataFeeder(place=place, feed_list=[x])
diff --git a/python/paddle/fluid/tests/unittests/test_weight_decay.py b/python/paddle/fluid/tests/unittests/test_weight_decay.py
index e125b5876f9ae7e635c6b7677b9d7164f26eb0f7..9a36be6173b3833037d678a1d72723a132fbfd1f 100644
--- a/python/paddle/fluid/tests/unittests/test_weight_decay.py
+++ b/python/paddle/fluid/tests/unittests/test_weight_decay.py
@@ -147,10 +147,12 @@ class TestWeightDecay(unittest.TestCase):
         startup_prog = fluid.framework.Program()
         startup_prog.random_seed = 1
         with prog_scope_guard(main_prog=main_prog, startup_prog=startup_prog):
-            data = fluid.layers.data(
-                name="words", shape=[1], dtype="int64", lod_level=1
+            data = paddle.static.data(
+                name="words", shape=[-1, 1], dtype="int64", lod_level=1
+            )
+            label = paddle.static.data(
+                name="label", shape=[-1, 1], dtype="int64"
             )
-            label = fluid.layers.data(name="label", shape=[1], dtype="int64")
             avg_cost = model(data, label, len(self.word_dict))
 
             param_list = [
diff --git a/python/paddle/fluid/tests/unittests/test_weight_normalization.py b/python/paddle/fluid/tests/unittests/test_weight_normalization.py
index e57e3ef9f0b767a7395e9794b42d3d37466f2b73..17a05bdb01caab7e930ad37e8c703e0ea4f557fe 100644
--- a/python/paddle/fluid/tests/unittests/test_weight_normalization.py
+++ b/python/paddle/fluid/tests/unittests/test_weight_normalization.py
@@ -35,8 +35,8 @@ class TestWeightNormalization(unittest.TestCase):
 
     @classmethod
     def set_program(cls):
-        data = fluid.layers.data(
-            name=cls.data_desc[0][0], shape=cls.data_desc[0][1]
+        data = paddle.static.data(
+            name=cls.data_desc[0][0], shape=[-1] + cls.data_desc[0][1]
         )
         out = paddle.static.nn.fc(
             x=data,
diff --git a/python/paddle/fluid/tests/unittests/test_where_op.py b/python/paddle/fluid/tests/unittests/test_where_op.py
index 34540f64e249c3ade8114e2f01e580a418e46cb9..32ee83614b7765b7c00ec405264651fc40e589e6 100644
--- a/python/paddle/fluid/tests/unittests/test_where_op.py
+++ b/python/paddle/fluid/tests/unittests/test_where_op.py
@@ -78,17 +78,22 @@ class TestWhereAPI(unittest.TestCase):
         for x_stop_gradient in [False, True]:
             for y_stop_gradient in [False, True]:
                 with fluid.program_guard(Program(), Program()):
-                    cond = fluid.layers.data(
-                        name='cond', shape=self.shape, dtype='bool'
+                    cond = paddle.static.data(
+                        name='cond', shape=[-1] + self.shape, dtype='bool'
                     )
-                    x = fluid.layers.data(
-                        name='x', shape=self.shape, dtype='float32'
+                    cond.desc.set_need_check_feed(False)
+                    x = paddle.static.data(
+                        name='x', shape=[-1] + self.shape, dtype='float32'
                     )
-                    y = fluid.layers.data(
-                        name='y', shape=self.shape, dtype='float32'
+                    x.desc.set_need_check_feed(False)
+                    y = paddle.static.data(
+                        name='y', shape=[-1] + self.shape, dtype='float32'
                     )
+                    y.desc.set_need_check_feed(False)
                     x.stop_gradient = x_stop_gradient
+                    x.desc.set_need_check_feed(False)
                     y.stop_gradient = y_stop_gradient
+                    y.desc.set_need_check_feed(False)
                     result = paddle.where(cond, x, y)
                     append_backward(paddle.mean(result))
                     for use_cuda in [False, True]:
@@ -127,8 +132,10 @@ class TestWhereAPI(unittest.TestCase):
     def test_api_broadcast(self, use_cuda=False):
         main_program = Program()
         with fluid.program_guard(main_program):
-            x = fluid.layers.data(name='x', shape=[4, 1], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[4, 2], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 4, 1], dtype='float32')
+            x.desc.set_need_check_feed(False)
+            y = paddle.static.data(name='y', shape=[-1, 4, 2], dtype='float32')
+            y.desc.set_need_check_feed(False)
             x_i = np.array([[0.9383, 0.1983, 3.2, 1.2]]).astype('float32')
             y_i = np.array([[1.0, 1.0, 1.0, 1.0], [1.0, 1.0, 1.0, 1.0]]).astype(
                 'float32'
@@ -151,9 +158,10 @@ class TestWhereAPI(unittest.TestCase):
         main_program = Program()
         with fluid.program_guard(main_program):
             cond_shape = [2, 4]
-            cond = fluid.layers.data(
-                name='cond', shape=cond_shape, dtype='bool'
+            cond = paddle.static.data(
+                name='cond', shape=[-1] + cond_shape, dtype='bool'
             )
+            cond.desc.set_need_check_feed(False)
             x_data = 1.0
             y_data = 2.0
             cond_data = np.array([False, False, True, True]).astype('bool')
@@ -175,11 +183,18 @@ class TestWhereAPI(unittest.TestCase):
         paddle.enable_static()
         main_program = Program()
         with fluid.program_guard(main_program):
-            cond = fluid.layers.data(
-                name='cond', shape=cond_shape, dtype='bool'
+            cond = paddle.static.data(
+                name='cond', shape=[-1] + cond_shape, dtype='bool'
             )
-            x = fluid.layers.data(name='x', shape=x_shape, dtype='float32')
-            y = fluid.layers.data(name='y', shape=y_shape, dtype='float32')
+            x = paddle.static.data(
+                name='x', shape=[-1] + x_shape, dtype='float32'
+            )
+            y = paddle.static.data(
+                name='y', shape=[-1] + y_shape, dtype='float32'
+            )
+            x.desc.set_need_check_feed(False)
+            y.desc.set_need_check_feed(False)
+            cond.desc.set_need_check_feed(False)
             cond_data_tmp = np.random.random(size=cond_shape).astype('float32')
             cond_data = cond_data_tmp < 0.3
             x_data = np.random.random(size=x_shape).astype('float32')
@@ -330,7 +345,8 @@ class TestWhereDygraphAPI(unittest.TestCase):
     def test_where_condition(self):
         data = np.array([[True, False], [False, True]])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[(-1), 2])
+            x = paddle.static.data(name='x', shape=[(-1), 2], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = paddle.where(x)
             self.assertEqual(type(y), tuple)
             self.assertEqual(len(y), 2)
@@ -343,7 +359,8 @@ class TestWhereDygraphAPI(unittest.TestCase):
         np.testing.assert_allclose(expect_out, np.array(res), rtol=1e-05)
         data = np.array([True, True, False])
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[(-1)])
+            x = paddle.static.data(name='x', shape=[(-1)], dtype='float32')
+            x.desc.set_need_check_feed(False)
             y = paddle.where(x)
             self.assertEqual(type(y), tuple)
             self.assertEqual(len(y), 1)
@@ -369,9 +386,14 @@ class TestWhereOpError(unittest.TestCase):
             self.assertRaises(TypeError, test_Variable)
 
             def test_type():
-                x = fluid.layers.data(name='x', shape=[4], dtype='bool')
-                y = fluid.layers.data(name='y', shape=[4], dtype='float16')
-                cond = fluid.layers.data(name='cond', shape=[4], dtype='int32')
+                x = paddle.static.data(name='x', shape=[-1, 4], dtype='bool')
+                x.desc.set_need_check_feed(False)
+                y = paddle.static.data(name='y', shape=[-1, 4], dtype='float16')
+                y.desc.set_need_check_feed(False)
+                cond = paddle.static.data(
+                    name='cond', shape=[-1, 4], dtype='int32'
+                )
+                cond.desc.set_need_check_feed(False)
                 paddle.where(cond, x, y)
 
             self.assertRaises(TypeError, test_type)
diff --git a/python/paddle/fluid/tests/unittests/test_while_loop_op.py b/python/paddle/fluid/tests/unittests/test_while_loop_op.py
index c5c31ac5717606214dac397c5734c014f3e05e77..c0a5b6a8cb9d3c47e28349a08666910bd20ac232 100644
--- a/python/paddle/fluid/tests/unittests/test_while_loop_op.py
+++ b/python/paddle/fluid/tests/unittests/test_while_loop_op.py
@@ -608,7 +608,7 @@ class TestApiWhileLoopSliceInBody(unittest.TestCase):
         main_program = Program()
         startup_program = Program()
         with program_guard(main_program, startup_program):
-            x = fluid.layers.data(name='x', shape=[5], dtype='int32')
+            x = paddle.static.data(name='x', shape=[-1, 5], dtype='int32')
             z = fluid.layers.fill_constant([1], 'int32', 0)
             x_shape = paddle.shape(x)
             i = fluid.layers.fill_constant([1], 'int32', 0)
diff --git a/python/paddle/fluid/tests/unittests/test_while_op.py b/python/paddle/fluid/tests/unittests/test_while_op.py
index 64b1ad2125b565163d44f8d14f0dc685b1470788..a78dbe6b7ea70577bda55e61ac4887c8ede45a3f 100644
--- a/python/paddle/fluid/tests/unittests/test_while_op.py
+++ b/python/paddle/fluid/tests/unittests/test_while_op.py
@@ -28,15 +28,9 @@ paddle.enable_static()
 
 class TestWhileOp(unittest.TestCase):
     def simple_net(self):
-        d0 = layers.data(
-            "d0", shape=[10], append_batch_size=False, dtype='float32'
-        )
-        d1 = layers.data(
-            "d1", shape=[10], append_batch_size=False, dtype='float32'
-        )
-        d2 = layers.data(
-            "d2", shape=[10], append_batch_size=False, dtype='float32'
-        )
+        d0 = paddle.static.data("d0", shape=[10], dtype='float32')
+        d1 = paddle.static.data("d1", shape=[10], dtype='float32')
+        d2 = paddle.static.data("d2", shape=[10], dtype='float32')
         i = layers.zeros(shape=[1], dtype='int64')
         i.stop_gradient = True
         init = layers.zeros(shape=[10], dtype='float32')
@@ -151,8 +145,10 @@ class TestIgnoreVarNameInWhile(unittest.TestCase):
             i = i + 1
             return [i, ten, batch_info, origin_seq]
 
-        x = fluid.layers.data(name='x', shape=[-1, 1, 4])
-        y = fluid.layers.data(name='y', shape=[-1, 1, 1])
+        x = paddle.static.data(name='x', shape=[-1, 1, 4], dtype='float32')
+        y = paddle.static.data(name='y', shape=[-1, 1, 1], dtype='float32')
+        x.desc.set_need_check_feed(False)
+        y.desc.set_need_check_feed(False)
         temp = layers.concat(input=[x, y], axis=-1)
         i = layers.fill_constant(shape=[1], value=0, dtype='int32')
         num = layers.fill_constant(shape=[1], value=5, dtype='int32')
@@ -207,7 +203,7 @@ class TestOutputsMustExistsInputs(unittest.TestCase):
                 return s
 
             paddle.enable_static()
-            x = paddle.static.data(shape=[-1], name='x')
+            x = paddle.static.data(shape=[-1], name='x', dtype='float32')
             func(x)
         for op in main_program.block(0).ops:
             if op.type == "while":
diff --git a/python/paddle/fluid/tests/unittests/transformer_model.py b/python/paddle/fluid/tests/unittests/transformer_model.py
index f4f755c131c9dc1492c473aa7d62dae6c8e94aac..42436b6e242b445bfceee93996c03dac7f13233b 100644
--- a/python/paddle/fluid/tests/unittests/transformer_model.py
+++ b/python/paddle/fluid/tests/unittests/transformer_model.py
@@ -499,11 +499,13 @@ def build_inputs(max_length, n_head):
 
     all_inputs = []
     for name, shape, dtype in zip(names, shapes, dtypes):
-        all_inputs.append(
-            fluid.layers.data(
-                name=name, shape=shape, dtype=dtype, append_batch_size=False
-            )
+        data_input = paddle.static.data(
+            name=name,
+            shape=shape,
+            dtype=dtype,
         )
+        data_input.desc.set_need_check_feed(False)
+        all_inputs.append(data_input)
     return all_inputs
 
 
diff --git a/python/paddle/fluid/tests/unittests/xpu/collective_allgather_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/collective_allgather_op_xpu.py
index 55304b8b40b4387fd8903f85fb977579c6f15f04..5cd5a92f4b2c4862d1d334eb6923fb08bd352d2f 100644
--- a/python/paddle/fluid/tests/unittests/xpu/collective_allgather_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/collective_allgather_op_xpu.py
@@ -16,7 +16,6 @@ from test_collective_base_xpu import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,8 +29,8 @@ class TestCollectiveAllGather(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             toutdata = main_prog.current_block().create_var(
                 name="outofgather",
diff --git a/python/paddle/fluid/tests/unittests/xpu/collective_allreduce_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/collective_allreduce_op_xpu.py
index 32196b77d258f045ba071a026e3f0ca1261bfab9..54b60f76665e3740b818d1bc50f03f17097b01e8 100644
--- a/python/paddle/fluid/tests/unittests/xpu/collective_allreduce_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/collective_allreduce_op_xpu.py
@@ -18,7 +18,6 @@ from test_collective_base_xpu import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -31,8 +30,8 @@ class TestCollectiveAllReduce(TestCollectiveRunnerBase):
     def get_model(self, main_prog, startup_program):
         ring_id = 0
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             toutdata = main_prog.current_block().create_var(
                 name="outofreduce",
diff --git a/python/paddle/fluid/tests/unittests/xpu/collective_identity_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/collective_identity_op_xpu.py
index 5a17539cbfb88596556216c473f78f417e829a3b..8fea9d7a4ac0e5ff4831949376fbb3e2c2ff333a 100644
--- a/python/paddle/fluid/tests/unittests/xpu/collective_identity_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/collective_identity_op_xpu.py
@@ -16,7 +16,6 @@ from test_collective_base_xpu import TestCollectiveRunnerBase, runtime_main
 
 import paddle
 import paddle.fluid as fluid
-import paddle.fluid.layers as layers
 from paddle.fluid import core
 
 paddle.enable_static()
@@ -30,8 +29,8 @@ class TestCollectiveIdentity(TestCollectiveRunnerBase):
         ring_id = 0
         nranks = 2
         with fluid.program_guard(main_prog, startup_program):
-            tindata = layers.data(
-                name="tindata", shape=[10, 1000], dtype='float32'
+            tindata = paddle.static.data(
+                name="tindata", shape=[-1, 10, 1000], dtype='float32'
             )
             toutdata = main_prog.current_block().create_var(
                 name="outofgather",
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_adadelta_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_adadelta_op_xpu.py
index f503e3cd4f595a22d3181efeba24756b069a0633..4de36728504ac875bacfd4a135f02d7c1c7ab045 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_adadelta_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_adadelta_op_xpu.py
@@ -168,8 +168,12 @@ class XPUTestAdadelta(XPUOpTestWrapper):
             place = fluid.XPUPlace(0)
             main = fluid.Program()
             with fluid.program_guard(main):
-                x = fluid.layers.data(name='x', shape=[13], dtype=self.dtype)
-                y = fluid.layers.data(name='y', shape=[1], dtype=self.dtype)
+                x = paddle.static.data(
+                    name='x', shape=[-1, 13], dtype=self.dtype
+                )
+                y = paddle.static.data(
+                    name='y', shape=[-1, 1], dtype=self.dtype
+                )
                 y_predict = paddle.static.nn.fc(x, size=1, activation=None)
                 cost = paddle.nn.functional.square_error_cost(
                     input=y_predict, label=y
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_conv3d_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_conv3d_op_xpu.py
index 915fb249514a964523589c95245efed158cfa93c..1367b972dd26fc38fd628f9c7dc13c33a678be76 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_conv3d_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_conv3d_op_xpu.py
@@ -22,7 +22,6 @@ from op_test_xpu import XPUOpTest
 from xpu.get_test_cover_info import XPUOpTestWrapper, create_test_class
 
 import paddle
-import paddle.fluid as fluid
 
 
 def conv3d_forward_naive(
@@ -567,17 +566,15 @@ class XPUTestConv3DOp_v2(XPUOpTestWrapper):
 class TestConv3DAPI(unittest.TestCase):
     def test_api(self):
 
-        input_NDHWC = fluid.layers.data(
+        input_NDHWC = paddle.static.data(
             name="input_NDHWC",
             shape=[2, 5, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
-        input_NCDHW = fluid.layers.data(
+        input_NCDHW = paddle.static.data(
             name="input_NCDHW",
             shape=[2, 3, 5, 5, 3],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -650,10 +647,9 @@ class TestConv3DAPI(unittest.TestCase):
 
 class TestConv3DAPI_Error(unittest.TestCase):
     def test_api(self):
-        input = fluid.layers.data(
+        input = paddle.static.data(
             name="input",
             shape=[2, 5, 5, 5, 4],
-            append_batch_size=False,
             dtype="float32",
         )
 
@@ -736,10 +732,9 @@ class TestConv3DAPI_Error(unittest.TestCase):
         self.assertRaises(ValueError, run_5)
 
         # ValueError: channel dimmention
-        x = fluid.layers.data(
+        x = paddle.static.data(
             name="x",
             shape=[2, 5, 5, 5, -1],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_dropout_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_dropout_op_xpu.py
index 011dd8fb9d88ef8280115f5a843f0e56c6642b16..fca45412fa61773dd06fb1ea0edef57a609a5337 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_dropout_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_dropout_op_xpu.py
@@ -141,8 +141,8 @@ class XPUTestDropoutOp(XPUOpTestWrapper):
                 def test_dtype():
                     # the input dtype of dropout must be float16 or float32 or float64
                     # float16 only can be set on GPU place
-                    x2 = fluid.layers.data(
-                        name='x2', shape=[3, 4, 5, 6], dtype="int32"
+                    x2 = paddle.static.data(
+                        name='x2', shape=[-1, 3, 4, 5, 6], dtype="int32"
                     )
                     paddle.nn.functional.dropout(x2, p=0.5)
 
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_expand_as_v2_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_expand_as_v2_op_xpu.py
index 9f7d39ca346aad367e855886eda60756ec244c30..7302e4ebe7300b92a6421dc607b61299caae273c 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_expand_as_v2_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_expand_as_v2_op_xpu.py
@@ -134,14 +134,11 @@ class TestExpandAsV2API(unittest.TestCase):
     def test_api(self):
         input1 = np.random.random([12, 14]).astype("float32")
         input2 = np.random.random([2, 12, 14]).astype("float32")
-        x = fluid.layers.data(
-            name='x', shape=[12, 14], append_batch_size=False, dtype="float32"
-        )
+        x = paddle.static.data(name='x', shape=[-1, 12, 14], dtype="float32")
 
-        y = fluid.layers.data(
+        y = paddle.static.data(
             name='target_tensor',
             shape=[2, 12, 14],
-            append_batch_size=False,
             dtype="float32",
         )
 
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_expand_v2_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_expand_v2_op_xpu.py
index 07039053dc0c6a99ca29b18c2d940e96cb0feb7a..558b177d842cf0154953f0b767bb21d33a424331 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_expand_v2_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_expand_v2_op_xpu.py
@@ -197,18 +197,16 @@ class TestExpandV2API(unittest.TestCase):
     def test_static(self):
         with fluid.program_guard(fluid.Program(), fluid.Program()):
             input = np.random.random([12, 14]).astype("float32")
-            x = fluid.layers.data(
+            x = paddle.static.data(
                 name='x',
-                shape=[12, 14],
-                append_batch_size=False,
+                shape=[-1, 12, 14],
                 dtype="float32",
             )
 
             positive_2 = fluid.layers.fill_constant([1], "int32", 12)
-            expand_shape = fluid.layers.data(
+            expand_shape = paddle.static.data(
                 name="expand_shape",
-                shape=[2],
-                append_batch_size=False,
+                shape=[-1, 2],
                 dtype="int32",
             )
 
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_index_select_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_index_select_op_xpu.py
index 85818e5a6a4ccc4b1763c072558279bc505fa20e..eb16430e39354fcd9020ffa5e96af0bc415fc613 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_index_select_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_index_select_op_xpu.py
@@ -102,10 +102,8 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         # case 1:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
-            index = fluid.layers.data(
-                name='index', shape=[3], dtype='int32', append_batch_size=False
-            )
+            x = paddle.static.data(name='x', shape=[-1, 4])
+            index = paddle.static.data(name='index', shape=[3], dtype='int32')
             z = paddle.index_select(x, index, axis=1)
             exe = fluid.Executor(fluid.XPUPlace(0))
             (res,) = exe.run(
@@ -120,10 +118,8 @@ class TestIndexSelectAPI(unittest.TestCase):
 
         # case 2:
         with program_guard(Program(), Program()):
-            x = fluid.layers.data(name='x', shape=[-1, 4])
-            index = fluid.layers.data(
-                name='index', shape=[3], dtype='int32', append_batch_size=False
-            )
+            x = paddle.static.data(name='x', shape=[-1, 4])
+            index = paddle.static.data(name='index', shape=[3], dtype='int32')
             z = paddle.index_select(x, index)
             exe = fluid.Executor(fluid.XPUPlace(0))
             (res,) = exe.run(
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_lookup_table_v2_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_lookup_table_v2_op_xpu.py
index f0db43acb685b50fd05fd914738370426ceb47d4..94645bcf9b2a24ebf7fc14a2850a6c8880ac1e70 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_lookup_table_v2_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_lookup_table_v2_op_xpu.py
@@ -167,7 +167,7 @@ class TestLookupTableWithTensorIdsWIsSelectedRows(
 
 class TestLookupTableApi(unittest.TestCase):
     def test_api(self):
-        x = fluid.layers.data(name='x', shape=[20], dtype='int64')
+        x = paddle.static.data(name='x', shape=[-1, 20], dtype='int64')
         emb = fluid.embedding(input=x, size=[128, 64])
 
         place = paddle.XPUPlace(0)
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_mean_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_mean_op_xpu.py
index 3991862346bbae813f3d98fa8c5994b544031d3d..a13bea88b6a8068f83d1d58cdf52c56f743c15c9 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_mean_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_mean_op_xpu.py
@@ -17,11 +17,11 @@ import unittest
 
 import numpy as np
 
+import paddle
+
 sys.path.append("..")
 from op_test_xpu import XPUOpTest
 
-import paddle
-import paddle.fluid as fluid
 from paddle.fluid import Program, program_guard
 
 np.random.seed(10)
@@ -91,12 +91,12 @@ class TestMeanOpError(unittest.TestCase):
             input1 = 12
             self.assertRaises(TypeError, paddle.mean, input1)
             # The input dtype of mean_op must be float16, float32, float64.
-            input2 = fluid.layers.data(
-                name='input2', shape=[12, 10], dtype="int32"
+            input2 = paddle.static.data(
+                name='input2', shape=[-1, 12, 10], dtype="int32"
             )
             self.assertRaises(TypeError, paddle.mean, input2)
-            input3 = fluid.layers.data(
-                name='input3', shape=[4], dtype="float16"
+            input3 = paddle.static.data(
+                name='input3', shape=[-1, 4], dtype="float16"
             )
             paddle.nn.functional.softmax(input3)
 
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_one_hot_v2_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_one_hot_v2_op_xpu.py
index dce5e263b9545fcd712522514ec6a98da11c274b..0a5752989e821bb6b664b84d19f5daa0f6228e02 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_one_hot_v2_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_one_hot_v2_op_xpu.py
@@ -167,7 +167,7 @@ class TestOneHotOpApi(unittest.TestCase):
             )
 
     def _run(self, depth):
-        label = fluid.layers.data(name="label", shape=[1], dtype="int64")
+        label = paddle.static.data(name="label", shape=[-1, 1], dtype="int64")
         one_hot_label = fluid.one_hot(input=label, depth=depth)
 
         place = fluid.XPUPlace(0)
@@ -191,10 +191,9 @@ class BadInputTestOnehotV2(unittest.TestCase):
         with fluid.program_guard(fluid.Program()):
 
             def test_bad_x():
-                label = fluid.layers.data(
+                label = paddle.static.data(
                     name="label",
                     shape=[4],
-                    append_batch_size=False,
                     dtype="float32",
                 )
                 one_hot_label = fluid.one_hot(input=label, depth=4)
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_sequence_conv_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_sequence_conv_op_xpu.py
index c3e49d123241d76e7e7bb56be0876c0b0b74a69a..6c2fd3fc9fa6c1a5cfcbf97d963d232f45d8c580 100755
--- a/python/paddle/fluid/tests/unittests/xpu/test_sequence_conv_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_sequence_conv_op_xpu.py
@@ -433,7 +433,7 @@ class TestSeqConvApi(unittest.TestCase):
     def test_api(self):
         import paddle.fluid as fluid
 
-        x = fluid.layers.data('x', shape=[32], lod_level=1)
+        x = paddle.static.data('x', shape=[-1, 32], lod_level=1)
         y = fluid.layers.sequence_conv(
             input=x, num_filters=2, filter_size=3, padding_start=None
         )
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_where_index_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_where_index_xpu.py
index e2aa83126351137010f99a175ffd7c761643c6f1..c70f6f61f1390d1617c9a68f5008f24d92da970e 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_where_index_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_where_index_xpu.py
@@ -101,7 +101,7 @@ for stype in support_types:
 class TestWhereOpError(unittest.TestCase):
     def test_api(self):
         with program_guard(Program(), Program()):
-            cond = fluid.layers.data(name='cond', shape=[4], dtype='bool')
+            cond = paddle.static.data(name='cond', shape=[-1, 4], dtype='bool')
             result = paddle.nonzero(cond)
 
             exe = fluid.Executor(paddle.XPUPlace(0))
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_where_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_where_op_xpu.py
index 4a080ba70cd967068ec3bb49fc53f8ad21667c9e..45aa192d72796b2bcce296667c4d01056115b2ce 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_where_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_where_op_xpu.py
@@ -150,8 +150,8 @@ class TestXPUWhereAPI(unittest.TestCase):
         train_prog = fluid.Program()
         startup = fluid.Program()
         with fluid.program_guard(train_prog, startup):
-            x = fluid.layers.data(name='x', shape=[4, 1], dtype='float32')
-            y = fluid.layers.data(name='y', shape=[4, 2], dtype='float32')
+            x = paddle.static.data(name='x', shape=[-1, 4, 1], dtype='float32')
+            y = paddle.static.data(name='y', shape=[-1, 4, 2], dtype='float32')
             x_i = np.array([[0.9383, 0.1983, 3.2, 1.2]]).astype("float32")
             y_i = np.array([[1.0, 1.0, 1.0, 1.0], [1.0, 1.0, 1.0, 1.0]]).astype(
                 "float32"
diff --git a/python/paddle/fluid/tests/unittests/xpu/test_while_op_xpu.py b/python/paddle/fluid/tests/unittests/xpu/test_while_op_xpu.py
index 10e3a6fddcdae704c26d3a6ee5e041555402ae11..56b93e494523ffb3a52a19096ac583d9aa7c7439 100644
--- a/python/paddle/fluid/tests/unittests/xpu/test_while_op_xpu.py
+++ b/python/paddle/fluid/tests/unittests/xpu/test_while_op_xpu.py
@@ -27,15 +27,9 @@ paddle.enable_static()
 
 class TestWhileOp(unittest.TestCase):
     def simple_net(self):
-        d0 = layers.data(
-            "d0", shape=[10], append_batch_size=False, dtype='float32'
-        )
-        d1 = layers.data(
-            "d1", shape=[10], append_batch_size=False, dtype='float32'
-        )
-        d2 = layers.data(
-            "d2", shape=[10], append_batch_size=False, dtype='float32'
-        )
+        d0 = paddle.static.data("d0", shape=[10], dtype='float32')
+        d1 = paddle.static.data("d1", shape=[10], dtype='float32')
+        d2 = paddle.static.data("d2", shape=[10], dtype='float32')
         i = layers.zeros(shape=[1], dtype='int64')
         i.stop_gradient = True
         init = layers.zeros(shape=[10], dtype='float32')
diff --git a/tools/codestyle/test_docstring_checker.py b/tools/codestyle/test_docstring_checker.py
index 8e18fd07030a8d9f0543cb8b44dc089fc55fe19d..ddc0ed185db4c6897371331e4c69b5a2637ec078 100644
--- a/tools/codestyle/test_docstring_checker.py
+++ b/tools/codestyle/test_docstring_checker.py
@@ -218,7 +218,7 @@ def fc(input,
         ValueError: If rank of the input tensor is less than 2.
     Examples:
         .. code-block:: python
-            data = fluid.layers.data(name="data", shape=[32, 32], dtype="float32")
+            data = paddle.static.data(name="data", shape=[-1, 32, 32], dtype="float32")
             fc = paddle.static.nn.fc(x=data, size=1000, activation="tanh")
     """
     raise ValueError('A very specific bad thing happened.')
diff --git a/tools/infrt/fake_models/multi_fc.py b/tools/infrt/fake_models/multi_fc.py
index ded9f67bd34db7bbe79e1ed84c96ca86bf55266a..8e3987a35219dda1eb074d30652b99dbabe1fa58 100644
--- a/tools/infrt/fake_models/multi_fc.py
+++ b/tools/infrt/fake_models/multi_fc.py
@@ -21,8 +21,8 @@ size = 2
 num_layers = 4
 paddle.enable_static()
 
-a = fluid.layers.data(name="A", shape=[-1, size], dtype='float32')
-label = fluid.layers.data(name="label", shape=[size], dtype='float32')
+a = paddle.static.data(name="A", shape=[-1, size], dtype='float32')
+label = paddle.static.data(name="label", shape=[-1, size], dtype='float32')
 
 fc_out = paddle.static.nn.fc(
     x=a,