diff --git a/python/paddle/dataset/uci_housing.py b/python/paddle/dataset/uci_housing.py
index 744f9104c51f5f..e834e2526318a8 100644
--- a/python/paddle/dataset/uci_housing.py
+++ b/python/paddle/dataset/uci_housing.py
@@ -51,7 +51,7 @@
 UCI_TRAIN_DATA = None
 UCI_TEST_DATA = None
 
-FLUID_URL_MODEL = 'https://github.com/PaddlePaddle/book/raw/develop/01.fit_a_line/base/fit_a_line.base.tar'
+FLUID_URL_MODEL = 'https://github.com/PaddlePaddle/book/raw/develop/01.fit_a_line/base/fit_a_line.fluid.tar'
 FLUID_MD5_MODEL = '6e6dd637ccd5993961f68bfbde46090b'
 
 
@@ -152,7 +152,7 @@ def reader():
 
 def base_model():
     parameter_tar = paddle.dataset.common.download(
-        FLUID_URL_MODEL, 'uci_housing', FLUID_MD5_MODEL, 'fit_a_line.base.tar'
+        FLUID_URL_MODEL, 'uci_housing', FLUID_MD5_MODEL, 'fit_a_line.fluid.tar'
     )
 
     tar = tarfile.TarFile(parameter_tar, mode='r')
diff --git a/python/paddle/distributed/auto_parallel/static/auto_align_tool.py b/python/paddle/distributed/auto_parallel/static/auto_align_tool.py
index 1d25a264bf6218..d7d98f75d80f1a 100644
--- a/python/paddle/distributed/auto_parallel/static/auto_align_tool.py
+++ b/python/paddle/distributed/auto_parallel/static/auto_align_tool.py
@@ -21,6 +21,8 @@
 
 import paddle
 import paddle.distributed as dist
+from paddle.base import core
+from paddle.base.framework import Program
 from paddle.distributed.auto_parallel.static.converter import Converter
 from paddle.distributed.auto_parallel.static.dist_context import (
     get_default_distributed_context,
@@ -30,8 +32,6 @@
     is_forward_op,
     is_loss_op,
 )
-from paddle.base import core
-from paddle.base.framework import Program
 from paddle.static.io import deserialize_program
 
 _valid_types = [
diff --git a/python/paddle/distributed/auto_parallel/static/completion.py b/python/paddle/distributed/auto_parallel/static/completion.py
index cb64025d7db0fe..2e7396dd596af1 100644
--- a/python/paddle/distributed/auto_parallel/static/completion.py
+++ b/python/paddle/distributed/auto_parallel/static/completion.py
@@ -15,8 +15,8 @@
 import copy
 import logging
 
-from paddle.distributed.fleet.meta_optimizers.common import OpRole
 from paddle.base.core import get_spmd_rule  # noqa: F401
+from paddle.distributed.fleet.meta_optimizers.common import OpRole
 from paddle.framework import core
 
 from ..process_mesh import ProcessMesh, compute_compatible_process_mesh
diff --git a/python/paddle/distributed/auto_parallel/static/engine.py b/python/paddle/distributed/auto_parallel/static/engine.py
index 972b989801e0a9..70af9150893306 100644
--- a/python/paddle/distributed/auto_parallel/static/engine.py
+++ b/python/paddle/distributed/auto_parallel/static/engine.py
@@ -24,8 +24,8 @@
 import paddle
 import paddle.distributed.auto_parallel.static.utils as auto_utils
 from paddle import static, utils
-from paddle.distributed import fleet
 from paddle.base.executor import _to_name_str
+from paddle.distributed import fleet
 from paddle.framework import IrGraph
 from paddle.framework import _current_expected_place as _get_device
 from paddle.framework import core, in_dynamic_mode
diff --git a/python/paddle/distributed/auto_parallel/static/tuner/rule_based_tuner.py b/python/paddle/distributed/auto_parallel/static/tuner/rule_based_tuner.py
index bb5bb4bbc2ac8a..8e0c3855e477e7 100644
--- a/python/paddle/distributed/auto_parallel/static/tuner/rule_based_tuner.py
+++ b/python/paddle/distributed/auto_parallel/static/tuner/rule_based_tuner.py
@@ -26,6 +26,9 @@
 import numpy as np
 
 import paddle
+from paddle.base import program_guard
+from paddle.base.backward import append_backward
+from paddle.base.framework import Parameter, unique_name
 from paddle.distributed.auto_parallel.process_mesh import ProcessMesh
 from paddle.distributed.auto_parallel.static.cluster_v2 import DeviceMesh
 from paddle.distributed.auto_parallel.static.completion import Completer
@@ -48,9 +51,6 @@
     print_program_with_dist_attr,
 )
 from paddle.distributed.fleet.meta_optimizers.common import OpRole
-from paddle.base import program_guard
-from paddle.base.backward import append_backward
-from paddle.base.framework import Parameter, unique_name
 
 from ....utils.log_utils import get_logger
 from ..graph import Graph
diff --git a/python/paddle/distributed/communication/stream/all_gather.py b/python/paddle/distributed/communication/stream/all_gather.py
index 04e0a7d5361d1f..165bf9690b6f27 100644
--- a/python/paddle/distributed/communication/stream/all_gather.py
+++ b/python/paddle/distributed/communication/stream/all_gather.py
@@ -15,8 +15,8 @@
 import paddle
 import paddle.distributed as dist
 from paddle import framework
-from paddle.distributed.communication.group import _get_global_group
 from paddle.base import data_feeder
+from paddle.distributed.communication.group import _get_global_group
 
 
 def _all_gather_into_tensor_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/all_reduce.py b/python/paddle/distributed/communication/stream/all_reduce.py
index b10773006a3441..61ea16b769a7ad 100644
--- a/python/paddle/distributed/communication/stream/all_reduce.py
+++ b/python/paddle/distributed/communication/stream/all_reduce.py
@@ -13,12 +13,12 @@
 # limitations under the License.
 
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _warn_cur_rank_not_in_group,
 )
 from paddle.distributed.communication.reduce import ReduceOp, _get_reduce_op
-from paddle.base import data_feeder
 
 
 def _all_reduce_in_dygraph(tensor, op, group, sync_op, use_calc_stream):
diff --git a/python/paddle/distributed/communication/stream/all_to_all.py b/python/paddle/distributed/communication/stream/all_to_all.py
index 7089ec70fc55d3..656986270805c9 100644
--- a/python/paddle/distributed/communication/stream/all_to_all.py
+++ b/python/paddle/distributed/communication/stream/all_to_all.py
@@ -15,11 +15,11 @@
 import paddle
 import paddle.distributed as dist
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _warn_cur_rank_not_in_group,
 )
-from paddle.base import data_feeder
 
 
 def _all_to_all_tensor_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/broadcast.py b/python/paddle/distributed/communication/stream/broadcast.py
index 2a671ac7a87006..2f02012098ec9a 100644
--- a/python/paddle/distributed/communication/stream/broadcast.py
+++ b/python/paddle/distributed/communication/stream/broadcast.py
@@ -13,12 +13,12 @@
 # limitations under the License.
 
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _get_or_throw_group_rank,
     _warn_cur_rank_not_in_group,
 )
-from paddle.base import data_feeder
 
 
 def _broadcast_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/recv.py b/python/paddle/distributed/communication/stream/recv.py
index d77010e79e061e..cb93ce8a471558 100644
--- a/python/paddle/distributed/communication/stream/recv.py
+++ b/python/paddle/distributed/communication/stream/recv.py
@@ -13,12 +13,12 @@
 # limitations under the License.
 
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _get_or_throw_group_rank,
     _warn_cur_rank_not_in_group,
 )
-from paddle.base import data_feeder
 
 
 def _recv_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/reduce.py b/python/paddle/distributed/communication/stream/reduce.py
index feb5aff33d0062..93c69a701879e8 100644
--- a/python/paddle/distributed/communication/stream/reduce.py
+++ b/python/paddle/distributed/communication/stream/reduce.py
@@ -13,13 +13,13 @@
 # limitations under the License.
 
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _get_or_throw_group_rank,
     _warn_cur_rank_not_in_group,
 )
 from paddle.distributed.communication.reduce import ReduceOp, _get_reduce_op
-from paddle.base import data_feeder
 
 
 def _reduce_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/reduce_scatter.py b/python/paddle/distributed/communication/stream/reduce_scatter.py
index d071db82b50ec8..6b9d669469e8ec 100644
--- a/python/paddle/distributed/communication/stream/reduce_scatter.py
+++ b/python/paddle/distributed/communication/stream/reduce_scatter.py
@@ -15,12 +15,12 @@
 import paddle
 import paddle.distributed as dist
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _warn_cur_rank_not_in_group,
 )
 from paddle.distributed.communication.reduce import ReduceOp, _get_reduce_op
-from paddle.base import data_feeder
 
 
 def _reduce_scatter_tensor_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/scatter.py b/python/paddle/distributed/communication/stream/scatter.py
index fc3a115ff63e41..84a03f44bdf3e6 100644
--- a/python/paddle/distributed/communication/stream/scatter.py
+++ b/python/paddle/distributed/communication/stream/scatter.py
@@ -17,12 +17,12 @@
 import paddle
 import paddle.distributed as dist
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _get_or_throw_group_rank,
     _warn_cur_rank_not_in_group,
 )
-from paddle.base import data_feeder
 
 
 def _scatter_tensor_in_dygraph(
diff --git a/python/paddle/distributed/communication/stream/send.py b/python/paddle/distributed/communication/stream/send.py
index 16705f05bbfd4d..d90d180783c850 100644
--- a/python/paddle/distributed/communication/stream/send.py
+++ b/python/paddle/distributed/communication/stream/send.py
@@ -13,12 +13,12 @@
 # limitations under the License.
 
 from paddle import framework
+from paddle.base import data_feeder
 from paddle.distributed.communication.group import (
     _get_global_group,
     _get_or_throw_group_rank,
     _warn_cur_rank_not_in_group,
 )
-from paddle.base import data_feeder
 
 
 def _send_in_dygraph(
diff --git a/python/paddle/distributed/fleet/base/distributed_strategy.py b/python/paddle/distributed/fleet/base/distributed_strategy.py
index e4da1ce7e1e6bc..8c778e480f4fa6 100755
--- a/python/paddle/distributed/fleet/base/distributed_strategy.py
+++ b/python/paddle/distributed/fleet/base/distributed_strategy.py
@@ -19,10 +19,10 @@
 import google.protobuf.text_format
 
 import paddle
-from paddle.distributed.fleet.proto import distributed_strategy_pb2
-from paddle.distributed.fleet.utils.log_util import logger
 from paddle.base.framework import _global_flags
 from paddle.base.wrapped_decorator import wrap_decorator
+from paddle.distributed.fleet.proto import distributed_strategy_pb2
+from paddle.distributed.fleet.utils.log_util import logger
 
 __all__ = []
 
diff --git a/python/paddle/distributed/fleet/base/role_maker.py b/python/paddle/distributed/fleet/base/role_maker.py
index 7300791dbcb7a5..870f818d177a9c 100755
--- a/python/paddle/distributed/fleet/base/role_maker.py
+++ b/python/paddle/distributed/fleet/base/role_maker.py
@@ -20,10 +20,10 @@
 import numpy as np
 
 import paddle
+from paddle.base import core
 from paddle.distributed.fleet.base.private_helper_function import (
     wait_server_ready,
 )
-from paddle.base import core
 
 from ...backup_env import getenv_or_backup
 
diff --git a/python/paddle/distributed/fleet/layers/mpu/mp_layers.py b/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
index 5cdc653e4443b5..44f39caeaa9394 100644
--- a/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
+++ b/python/paddle/distributed/fleet/layers/mpu/mp_layers.py
@@ -14,8 +14,8 @@
 
 import paddle
 from paddle.autograd import PyLayer
-from paddle.distributed import fleet
 from paddle.base import core
+from paddle.distributed import fleet
 from paddle.nn import functional as F
 
 from ....communication.reduce import ReduceOp, _get_reduce_op
diff --git a/python/paddle/distributed/fleet/layers/mpu/mp_ops.py b/python/paddle/distributed/fleet/layers/mpu/mp_ops.py
index 4773f6e0b66e34..267652d3f9268b 100644
--- a/python/paddle/distributed/fleet/layers/mpu/mp_ops.py
+++ b/python/paddle/distributed/fleet/layers/mpu/mp_ops.py
@@ -15,8 +15,8 @@
 import paddle
 from paddle import _legacy_C_ops
 from paddle.autograd import PyLayer
-from paddle.distributed import collective
 from paddle.base.data_feeder import check_dtype, check_variable_and_dtype
+from paddle.distributed import collective
 from paddle.framework import LayerHelper, _create_tensor, in_dynamic_mode
 from paddle.nn import Layer
 from paddle.nn.utils import dygraph_utils
diff --git a/python/paddle/distributed/fleet/layers/mpu/random.py b/python/paddle/distributed/fleet/layers/mpu/random.py
index 22063aa24fa799..5b43ef951cfff4 100644
--- a/python/paddle/distributed/fleet/layers/mpu/random.py
+++ b/python/paddle/distributed/fleet/layers/mpu/random.py
@@ -18,9 +18,9 @@
 
 import paddle
 from paddle import _legacy_C_ops
-from paddle.common_ops_import import Variable
 from paddle.base import core
 from paddle.base.data_feeder import check_variable_and_dtype
+from paddle.common_ops_import import Variable
 from paddle.framework import LayerHelper, in_dynamic_mode
 
 __all__ = []
diff --git a/python/paddle/distributed/fleet/meta_optimizers/dgc_optimizer.py b/python/paddle/distributed/fleet/meta_optimizers/dgc_optimizer.py
index 4f74b62a418a55..cee43657c75275 100644
--- a/python/paddle/distributed/fleet/meta_optimizers/dgc_optimizer.py
+++ b/python/paddle/distributed/fleet/meta_optimizers/dgc_optimizer.py
@@ -19,9 +19,9 @@
 __all__ = []
 
 import paddle
-from paddle.common_ops_import import LayerHelper
 from paddle.base import framework
 from paddle.base.dygraph import base as imperative_base
+from paddle.common_ops_import import LayerHelper
 from paddle.framework import core, in_dynamic_mode
 from paddle.nn.clip import ClipGradByNorm, append_gradient_clip_ops
 from paddle.optimizer import Momentum, Optimizer
diff --git a/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_stage3.py b/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_stage3.py
index 3ef5dbb91dd133..3b342778442c26 100644
--- a/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_stage3.py
+++ b/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_stage3.py
@@ -22,8 +22,8 @@
 import paddle.distributed as dist
 from paddle import framework, nn
 from paddle.autograd import PyLayer
-from paddle.distributed import collective
 from paddle.base.framework import EagerParamBase
+from paddle.distributed import collective
 from paddle.framework import core
 from paddle.nn import ClipGradByGlobalNorm
 
diff --git a/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_utils.py b/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_utils.py
index 3e4ba5026b1fa5..adfad3ec77e319 100644
--- a/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_utils.py
+++ b/python/paddle/distributed/fleet/meta_parallel/sharding/group_sharded_utils.py
@@ -20,9 +20,9 @@
 
 import paddle
 from paddle import _C_ops, _legacy_C_ops
-from paddle.common_ops_import import dygraph_only
 from paddle.base import core
 from paddle.base.dygraph import to_variable
+from paddle.common_ops_import import dygraph_only
 from paddle.nn import clip
 
 
diff --git a/python/paddle/distributed/fleet/recompute/recompute_hybrid.py b/python/paddle/distributed/fleet/recompute/recompute_hybrid.py
index 5a8ca327ffc513..bdbdc797b2b7b1 100644
--- a/python/paddle/distributed/fleet/recompute/recompute_hybrid.py
+++ b/python/paddle/distributed/fleet/recompute/recompute_hybrid.py
@@ -158,7 +158,7 @@ def forward(
                 # Note:
                 #  If not marked non_differentiable, all output tensors' attr `stop gradient`
                 #  will be reset to `False` in c++ backend.
-                #  See https://github.com/PaddlePaddle/Paddle/blob/9d62efb0e6e5373823039d9eda96cd5905426c0a/paddle/base/pybind/eager_py_layer.cc#L388
+                #  See https://github.com/PaddlePaddle/Paddle/blob/9d62efb0e6e5373823039d9eda96cd5905426c0a/paddle/fluid/pybind/eager_py_layer.cc#L388
                 if framework.in_dynamic_mode() and state:
                     ctx.mark_non_differentiable(arg)
             else:
diff --git a/python/paddle/distributed/fleet/scaler.py b/python/paddle/distributed/fleet/scaler.py
index a9528210c62231..bf0d7363b05251 100755
--- a/python/paddle/distributed/fleet/scaler.py
+++ b/python/paddle/distributed/fleet/scaler.py
@@ -18,8 +18,8 @@
 
 import paddle
 from paddle import _C_ops, _legacy_C_ops
-from paddle.distributed import fleet
 from paddle.base.dygraph import to_variable
+from paddle.distributed import fleet
 from paddle.framework import core
 
 from .base.topology import ParallelMode
diff --git a/python/paddle/distributed/fleet/utils/hybrid_parallel_inference.py b/python/paddle/distributed/fleet/utils/hybrid_parallel_inference.py
index ae9fccb1be7a6b..a5723f856e6613 100644
--- a/python/paddle/distributed/fleet/utils/hybrid_parallel_inference.py
+++ b/python/paddle/distributed/fleet/utils/hybrid_parallel_inference.py
@@ -16,10 +16,9 @@
 
 import numpy as np
 
-from paddle.distributed import fleet
-
 # (TODO: GhostScreaming) It will be removed later.
 from paddle.base import core
+from paddle.distributed import fleet
 from paddle.framework import Block, Program, in_dynamic_mode
 
 
diff --git a/python/paddle/distributed/fleet/utils/hybrid_parallel_util.py b/python/paddle/distributed/fleet/utils/hybrid_parallel_util.py
index f44ba3f41ba053..edeb585384b03f 100644
--- a/python/paddle/distributed/fleet/utils/hybrid_parallel_util.py
+++ b/python/paddle/distributed/fleet/utils/hybrid_parallel_util.py
@@ -14,6 +14,9 @@
 
 import paddle
 from paddle import framework
+
+# (TODO: GhostScreaming) It will be removed later.
+from paddle.base import core
 from paddle.distributed.parallel import (
     _split_tensors,
     build_groups,
@@ -21,9 +24,6 @@
     sync_params_buffers,
 )
 
-# (TODO: GhostScreaming) It will be removed later.
-from paddle.base import core
-
 from .log_util import logger
 
 __all__ = []
diff --git a/python/paddle/distributed/fleet/utils/mix_precision_utils.py b/python/paddle/distributed/fleet/utils/mix_precision_utils.py
index 37e6a4d32a42e9..ead12379bb5a86 100644
--- a/python/paddle/distributed/fleet/utils/mix_precision_utils.py
+++ b/python/paddle/distributed/fleet/utils/mix_precision_utils.py
@@ -20,13 +20,13 @@
 
 import paddle
 from paddle import _legacy_C_ops, nn
+from paddle.base import framework
+from paddle.base.dygraph import base as imperative_base
+from paddle.base.dygraph import to_variable
 from paddle.distributed import fleet
 from paddle.distributed.fleet.utils.hybrid_parallel_util import (
     obtain_optimizer_parameters_list,
 )
-from paddle.base import framework
-from paddle.base.dygraph import base as imperative_base
-from paddle.base.dygraph import to_variable
 from paddle.framework import core
 
 
diff --git a/python/paddle/distributed/fleet/utils/sequence_parallel_utils.py b/python/paddle/distributed/fleet/utils/sequence_parallel_utils.py
index c43a339db9f3fe..ae5dec21b4e8ff 100644
--- a/python/paddle/distributed/fleet/utils/sequence_parallel_utils.py
+++ b/python/paddle/distributed/fleet/utils/sequence_parallel_utils.py
@@ -17,12 +17,12 @@
 import paddle
 from paddle import distributed as dist
 from paddle.autograd import PyLayer
+from paddle.base import core
 from paddle.distributed import fleet
 from paddle.distributed.fleet.meta_parallel import get_rng_state_tracker
 from paddle.distributed.fleet.utils.hybrid_parallel_util import (
     fused_allreduce_gradients_with_group,
 )
-from paddle.base import core
 from paddle.nn import Layer
 from paddle.nn import functional as F
 
diff --git a/python/paddle/distributed/fleet/utils/tensor_parallel_utils.py b/python/paddle/distributed/fleet/utils/tensor_parallel_utils.py
index 0ad0c5024015ec..e370042927434f 100644
--- a/python/paddle/distributed/fleet/utils/tensor_parallel_utils.py
+++ b/python/paddle/distributed/fleet/utils/tensor_parallel_utils.py
@@ -22,8 +22,8 @@
 ch.setFormatter(formatter)
 logger.addHandler(ch)
 
-from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY
 from paddle.base import core
+from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY
 from paddle.static import Parameter
 
 _supported_optimizer_type = [
diff --git a/python/paddle/distributed/launch/context/device.py b/python/paddle/distributed/launch/context/device.py
index dfba7b72149fb9..5126f687ccb0a5 100644
--- a/python/paddle/distributed/launch/context/device.py
+++ b/python/paddle/distributed/launch/context/device.py
@@ -14,10 +14,9 @@
 
 import os
 
-from paddle.device import get_available_custom_device
-
 # (TODO: GhostScreaming) It will be removed later.
 from paddle.base import core
+from paddle.device import get_available_custom_device
 
 
 class DeviceType:
diff --git a/python/paddle/distributed/parallel_with_gloo.py b/python/paddle/distributed/parallel_with_gloo.py
index 1a4bf4f8fbe6ee..9183139becd95f 100755
--- a/python/paddle/distributed/parallel_with_gloo.py
+++ b/python/paddle/distributed/parallel_with_gloo.py
@@ -15,13 +15,12 @@
 import time
 from multiprocessing import Manager, Process
 
-from paddle.distributed.fleet.base.private_helper_function import (
-    wait_server_ready,
-)
-
 # deprecated module import
 # (TODO: GhostScreaming) It will be removed later.
 from paddle.base import core
+from paddle.distributed.fleet.base.private_helper_function import (
+    wait_server_ready,
+)
 
 __all__ = []
 
diff --git a/python/paddle/distributed/passes/auto_parallel_amp.py b/python/paddle/distributed/passes/auto_parallel_amp.py
index 9dc4f15f65a284..2f563cba7e72ac 100644
--- a/python/paddle/distributed/passes/auto_parallel_amp.py
+++ b/python/paddle/distributed/passes/auto_parallel_amp.py
@@ -13,6 +13,7 @@
 # limitations under the License.
 
 import paddle
+from paddle.base.data_feeder import check_type, check_variable_and_dtype
 from paddle.distributed.auto_parallel.static.dist_attribute import (
     OperatorDistAttr,
 )
@@ -24,7 +25,6 @@
     set_var_dist_attr,
 )
 from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY, OpRole
-from paddle.base.data_feeder import check_type, check_variable_and_dtype
 from paddle.framework import core
 from paddle.static.amp.bf16.amp_utils import (
     AutoMixedPrecisionListsBF16,
diff --git a/python/paddle/distributed/passes/auto_parallel_pipeline.py b/python/paddle/distributed/passes/auto_parallel_pipeline.py
index de7e819635a490..9e2a06778854aa 100644
--- a/python/paddle/distributed/passes/auto_parallel_pipeline.py
+++ b/python/paddle/distributed/passes/auto_parallel_pipeline.py
@@ -14,6 +14,8 @@
 
 import os
 
+from paddle.base import core
+from paddle.base.framework import Program
 from paddle.distributed.auto_parallel.static.process_group import (
     remove_process_group,
 )
@@ -24,8 +26,6 @@
     is_optimize_op,
 )
 from paddle.distributed.fleet.fleet_executor_utils import TaskNode
-from paddle.base import core
-from paddle.base.framework import Program
 
 from .pass_base import PassBase, register_pass
 from .pass_utils import _create_program, _insert_sync_for_fthenb_1f1b
diff --git a/python/paddle/distributed/passes/auto_parallel_recompute.py b/python/paddle/distributed/passes/auto_parallel_recompute.py
index 194aeefb6029bd..0fd008ff5a7012 100644
--- a/python/paddle/distributed/passes/auto_parallel_recompute.py
+++ b/python/paddle/distributed/passes/auto_parallel_recompute.py
@@ -15,7 +15,6 @@
 import logging
 
 import paddle
-from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY, OpRole
 from paddle.base.backward import (
     ProgramStats,
     _append_grad_suffix_,
@@ -23,6 +22,7 @@
     _get_no_grad_set_name,
     _rename_arg_,
 )
+from paddle.distributed.fleet.meta_optimizers.common import OP_ROLE_KEY, OpRole
 from paddle.framework import core
 from paddle.utils import unique_name
 
diff --git a/python/paddle/distributed/passes/pass_utils.py b/python/paddle/distributed/passes/pass_utils.py
index 317b43c34bed97..b353d91cb1c88d 100644
--- a/python/paddle/distributed/passes/pass_utils.py
+++ b/python/paddle/distributed/passes/pass_utils.py
@@ -15,6 +15,8 @@
 from collections import OrderedDict
 from typing import List
 
+from paddle.base import core
+from paddle.base.framework import Parameter, Program
 from paddle.distributed.auto_parallel.static.utils import (
     is_backward_op,
     is_forward_op,
@@ -22,8 +24,6 @@
     is_optimize_op,
 )
 from paddle.distributed.fleet.meta_optimizers.common import OpRole
-from paddle.base import core
-from paddle.base.framework import Parameter, Program
 
 __not_shape_var_type__ = [
     core.VarDesc.VarType.READER,
diff --git a/python/paddle/distributed/passes/pipeline_pass_base.py b/python/paddle/distributed/passes/pipeline_pass_base.py
index 65e09a73327d36..07235ecc05a0f8 100644
--- a/python/paddle/distributed/passes/pipeline_pass_base.py
+++ b/python/paddle/distributed/passes/pipeline_pass_base.py
@@ -14,8 +14,8 @@
 
 import logging
 
-from paddle.distributed.auto_parallel.static.utils import get_logger
 from paddle.base import core
+from paddle.distributed.auto_parallel.static.utils import get_logger
 
 from .pass_base import PassBase
 from .pass_utils import get_skip_gc_vars
diff --git a/python/paddle/distributed/passes/ps_trainer_pass.py b/python/paddle/distributed/passes/ps_trainer_pass.py
index 85dc873b7112fc..434d32891d1d00 100755
--- a/python/paddle/distributed/passes/ps_trainer_pass.py
+++ b/python/paddle/distributed/passes/ps_trainer_pass.py
@@ -17,8 +17,8 @@
 from _collections import defaultdict
 
 import paddle
-from paddle.distributed.passes.pass_base import PassBase, register_pass
 from paddle.base import framework
+from paddle.distributed.passes.pass_base import PassBase, register_pass
 from paddle.framework import core
 from paddle.static import Parameter, Program
 
diff --git a/python/paddle/distributed/ps/utils/collective_transpiler.py b/python/paddle/distributed/ps/utils/collective_transpiler.py
index 989a97d1d8426a..99bb76a3b315b3 100644
--- a/python/paddle/distributed/ps/utils/collective_transpiler.py
+++ b/python/paddle/distributed/ps/utils/collective_transpiler.py
@@ -15,10 +15,10 @@
 import os
 
 import paddle
+from paddle.base import unique_name
 from paddle.distributed.fleet.base.private_helper_function import (
     wait_server_ready,
 )
-from paddle.base import unique_name
 from paddle.framework import core
 from paddle.static import default_main_program, default_startup_program
 
diff --git a/python/paddle/distributed/ps/utils/ps_program_builder.py b/python/paddle/distributed/ps/utils/ps_program_builder.py
index 428875fff6f72b..8ba98a3707d99c 100755
--- a/python/paddle/distributed/ps/utils/ps_program_builder.py
+++ b/python/paddle/distributed/ps/utils/ps_program_builder.py
@@ -97,9 +97,7 @@ def _build_programs(self):
         elif self.attrs['is_server']:
             self._build_pserver_programs()
             self.loss.block.program = self.attrs['_main_server']
-            base.framework.switch_startup_program(
-                self.attrs['_startup_server']
-            )
+            base.framework.switch_startup_program(self.attrs['_startup_server'])
 
 
 class GeoPsProgramBuilder(PsProgramBuilder):  # 仅 CPU 模式
@@ -372,9 +370,7 @@ def _build_programs(self):
         elif self.attrs['is_server']:
             self._build_pserver_programs()
             self.loss.block.program = self.attrs['_main_server']
-            base.framework.switch_startup_program(
-                self.attrs['_startup_server']
-            )
+            base.framework.switch_startup_program(self.attrs['_startup_server'])
 
 
 class FlPsProgramBuilder(HeterAsyncPsProgramBuilder):
@@ -479,7 +475,5 @@ def _build_programs(self):
             )
         else:
             self._build_pserver_programs()
-            base.framework.switch_startup_program(
-                self.attrs['_startup_server']
-            )
+            base.framework.switch_startup_program(self.attrs['_startup_server'])
             paddle.framework.switch_main_program(self.attrs['_main_server'])
diff --git a/python/paddle/distributed/ps/utils/public.py b/python/paddle/distributed/ps/utils/public.py
index 82d6eccee2e10f..865de4c828308b 100755
--- a/python/paddle/distributed/ps/utils/public.py
+++ b/python/paddle/distributed/ps/utils/public.py
@@ -18,8 +18,8 @@
 import warnings
 from functools import reduce
 
-from paddle.distributed.io import is_persistable
 from paddle.base.framework import generate_control_dev_var_name
+from paddle.distributed.io import is_persistable
 from paddle.framework import core
 
 # logging.basicConfig(
diff --git a/python/paddle/distributed/rpc/rpc.py b/python/paddle/distributed/rpc/rpc.py
index ae4c1459d8717b..ebe6bc54623d6f 100644
--- a/python/paddle/distributed/rpc/rpc.py
+++ b/python/paddle/distributed/rpc/rpc.py
@@ -18,10 +18,10 @@
 import time
 from collections import namedtuple
 
+from paddle.base import core
 from paddle.distributed.launch.context import Node
 from paddle.distributed.rpc.internal import PythonFunc, _serialize
 from paddle.distributed.utils.launch_utils import logger
-from paddle.base import core
 
 WorkerInfo = namedtuple("WorkerInfo", ["name", "rank", "ip", "port"])
 
diff --git a/python/paddle/distributed/spawn.py b/python/paddle/distributed/spawn.py
index c3351c5c9436b7..91039b3b3bac3b 100644
--- a/python/paddle/distributed/spawn.py
+++ b/python/paddle/distributed/spawn.py
@@ -18,6 +18,9 @@
 import sys
 import warnings
 
+# deprecated module import
+# (TODO: GhostScreaming) It will be removed later.
+from paddle.base import core
 from paddle.device import get_device
 from paddle.distributed.cloud_utils import (
     _get_trainers_num,
@@ -35,10 +38,6 @@
     _print_arguments,
     get_host_name_ip,
 )
-
-# deprecated module import
-# (TODO: GhostScreaming) It will be removed later.
-from paddle.base import core
 from paddle.framework import set_flags
 
 __all__ = []
diff --git a/python/paddle/distributed/transpiler/collective.py b/python/paddle/distributed/transpiler/collective.py
index 03b3be36e69b0c..e76238e02af43e 100644
--- a/python/paddle/distributed/transpiler/collective.py
+++ b/python/paddle/distributed/transpiler/collective.py
@@ -15,10 +15,10 @@
 import os
 
 import paddle
+from paddle.base import unique_name
 from paddle.distributed.fleet.base.private_helper_function import (
     wait_server_ready,
 )
-from paddle.base import unique_name
 from paddle.framework import core
 from paddle.static import default_main_program, default_startup_program
 
diff --git a/python/paddle/distribution/bernoulli.py b/python/paddle/distribution/bernoulli.py
index c86bed5a0d78ff..7d4849fab48e7c 100644
--- a/python/paddle/distribution/bernoulli.py
+++ b/python/paddle/distribution/bernoulli.py
@@ -16,9 +16,9 @@
 import numpy as np
 
 import paddle
-from paddle.distribution import exponential_family
 from paddle.base.data_feeder import check_type, convert_dtype
 from paddle.base.framework import Variable
+from paddle.distribution import exponential_family
 from paddle.framework import in_dynamic_mode
 from paddle.nn.functional import (
     binary_cross_entropy_with_logits,
diff --git a/python/paddle/distribution/categorical.py b/python/paddle/distribution/categorical.py
index 110f324db77c71..b6484e3f21d563 100644
--- a/python/paddle/distribution/categorical.py
+++ b/python/paddle/distribution/categorical.py
@@ -15,9 +15,9 @@
 import numpy as np
 
 import paddle
-from paddle.distribution import distribution
 from paddle.base.data_feeder import check_type, convert_dtype
 from paddle.base.framework import Variable
+from paddle.distribution import distribution
 from paddle.framework import in_dynamic_mode
 from paddle.tensor import multinomial
 
diff --git a/python/paddle/distribution/cauchy.py b/python/paddle/distribution/cauchy.py
index cad5c88753421e..63e0a5f252638d 100644
--- a/python/paddle/distribution/cauchy.py
+++ b/python/paddle/distribution/cauchy.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distribution import distribution
 from paddle.base import framework
+from paddle.distribution import distribution
 
 
 class Cauchy(distribution.Distribution):
diff --git a/python/paddle/distribution/dirichlet.py b/python/paddle/distribution/dirichlet.py
index a1695f2b36b485..cf578c9d0dd5c9 100644
--- a/python/paddle/distribution/dirichlet.py
+++ b/python/paddle/distribution/dirichlet.py
@@ -13,9 +13,9 @@
 # limitations under the License.
 
 import paddle
-from paddle.distribution import exponential_family
 from paddle.base.data_feeder import check_variable_and_dtype
 from paddle.base.layer_helper import LayerHelper
+from paddle.distribution import exponential_family
 from paddle.framework import in_dynamic_mode
 
 
diff --git a/python/paddle/distribution/geometric.py b/python/paddle/distribution/geometric.py
index e4f2795b31f160..bfcd734bb17850 100644
--- a/python/paddle/distribution/geometric.py
+++ b/python/paddle/distribution/geometric.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distribution import distribution, uniform
 from paddle.base import framework
+from paddle.distribution import distribution, uniform
 
 
 class Geometric(distribution.Distribution):
diff --git a/python/paddle/distribution/gumbel.py b/python/paddle/distribution/gumbel.py
index 6c21b8d601a3f8..005801ae6b7ccc 100644
--- a/python/paddle/distribution/gumbel.py
+++ b/python/paddle/distribution/gumbel.py
@@ -18,8 +18,8 @@
 import numpy as np
 
 import paddle
-from paddle.distribution.transformed_distribution import TransformedDistribution
 from paddle.base import framework
+from paddle.distribution.transformed_distribution import TransformedDistribution
 
 
 class Gumbel(TransformedDistribution):
diff --git a/python/paddle/distribution/laplace.py b/python/paddle/distribution/laplace.py
index 3d5ef1bbf245a9..fc4b57eeba79c6 100644
--- a/python/paddle/distribution/laplace.py
+++ b/python/paddle/distribution/laplace.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distribution import distribution
 from paddle.base import framework
+from paddle.distribution import distribution
 
 
 class Laplace(distribution.Distribution):
diff --git a/python/paddle/distribution/normal.py b/python/paddle/distribution/normal.py
index f522ac65c8a6d8..0564bb2b5397be 100644
--- a/python/paddle/distribution/normal.py
+++ b/python/paddle/distribution/normal.py
@@ -18,9 +18,9 @@
 import numpy as np
 
 import paddle
-from paddle.distribution import distribution
 from paddle.base.data_feeder import check_type, convert_dtype
 from paddle.base.framework import Variable
+from paddle.distribution import distribution
 from paddle.framework import in_dynamic_mode
 from paddle.tensor import random
 
diff --git a/python/paddle/distribution/uniform.py b/python/paddle/distribution/uniform.py
index 833195491e0380..4d8a0f97d910dd 100644
--- a/python/paddle/distribution/uniform.py
+++ b/python/paddle/distribution/uniform.py
@@ -16,9 +16,9 @@
 
 import paddle
 from paddle import _C_ops
-from paddle.distribution import distribution
 from paddle.base.data_feeder import check_type, convert_dtype
 from paddle.base.framework import Variable
+from paddle.distribution import distribution
 from paddle.framework import in_dynamic_mode
 from paddle.tensor import random
 
diff --git a/python/paddle/hapi/model.py b/python/paddle/hapi/model.py
index 9002de1ee1698c..8ff1a557673af3 100644
--- a/python/paddle/hapi/model.py
+++ b/python/paddle/hapi/model.py
@@ -26,14 +26,14 @@
 import paddle.distributed as dist
 from paddle import base
 from paddle.autograd import no_grad
-from paddle.distributed import fleet
-from paddle.distributed.fleet.base import role_maker
 from paddle.base import core
 from paddle.base.dygraph.base import to_variable
 from paddle.base.executor import global_scope
 from paddle.base.framework import Variable
 from paddle.base.framework import _current_expected_place as _get_device
 from paddle.base.framework import _get_paddle_place
+from paddle.distributed import fleet
+from paddle.distributed.fleet.base import role_maker
 from paddle.framework import in_dynamic_mode
 from paddle.framework.io_utils import is_belong_to_optimizer
 from paddle.io import DataLoader, Dataset, DistributedBatchSampler
@@ -58,9 +58,7 @@ def to_list(value):
 
 
 def to_numpy(var):
-    assert isinstance(
-        var, (Variable, base.core.eager.Tensor)
-    ), "not a variable"
+    assert isinstance(var, (Variable, base.core.eager.Tensor)), "not a variable"
     if isinstance(var, base.core.eager.Tensor):
         return np.array(var)
     t = global_scope().find_var(var.name).get_tensor()
diff --git a/python/paddle/hapi/model_summary.py b/python/paddle/hapi/model_summary.py
index 4a204f64b3d63e..267e938a6e298a 100644
--- a/python/paddle/hapi/model_summary.py
+++ b/python/paddle/hapi/model_summary.py
@@ -353,9 +353,7 @@ def _build_dtypes(input_size, dtype):
     depth = len(list(model.sublayers()))
 
     def _get_shape_from_tensor(x):
-        if isinstance(
-            x, (paddle.base.Variable, paddle.base.core.eager.Tensor)
-        ):
+        if isinstance(x, (paddle.base.Variable, paddle.base.core.eager.Tensor)):
             return list(x.shape)
         elif isinstance(x, (list, tuple)):
             return [_get_shape_from_tensor(xx) for xx in x]
diff --git a/python/paddle/incubate/autograd/primreg.py b/python/paddle/incubate/autograd/primreg.py
index 5e94c220acb471..5cedac46320ddc 100644
--- a/python/paddle/incubate/autograd/primreg.py
+++ b/python/paddle/incubate/autograd/primreg.py
@@ -79,7 +79,7 @@ def op_position_inputs(op):
         .. code-block:: python
 
             >>> from paddle.incubate.autograd.primops import _simple_binop
-            >>> from paddle.fluid.layer_helper import LayerHelper
+            >>> from paddle.base.layer_helper import LayerHelper
             >>> from paddle.incubate.autograd.primreg import REGISTER_FN
 
             >>> # doctest: +SKIP('Depends on external code.')
@@ -126,7 +126,7 @@ def op_position_output(op):
 
             >>> # doctest: +SKIP('Depends on external code.')
             >>> from paddle.incubate.autograd.primops import _simple_binop
-            >>> from paddle.fluid.layer_helper import LayerHelper
+            >>> from paddle.base.layer_helper import LayerHelper
             >>> from paddle.incubate.autograd.primreg import REGISTER_FN
 
             >>> @REGISTER_FN('div_p', 'X', 'Y', 'Z')
@@ -169,7 +169,7 @@ def REGISTER_FN(op_type, *position_argnames):
 
             >>> # doctest: +SKIP('Depends on external code.')
             >>> from paddle.incubate.autograd.primops import _simple_binop
-            >>> from paddle.fluid.layer_helper import LayerHelper
+            >>> from paddle.base.layer_helper import LayerHelper
             >>> from paddle.incubate.autograd.primreg import REGISTER_FN
 
             >>> @REGISTER_FN('tanh_p', 'X', 'Y')
@@ -204,7 +204,7 @@ def REGISTER_ORIG2PRIM(op_type):
         .. code-block:: python
 
             >>> # doctest: +SKIP('Depends on external code.')
-            >>> from paddle.fluid.layer_helper import LayerHelper
+            >>> from paddle.base.layer_helper import LayerHelper
             >>> from paddle.incubate.autograd.utils import get_input_var_list
             >>> from paddle.incubate.autograd import primops
             >>> from paddle.incubate.autograd.primreg import REGISTER_ORIG2PRIM
diff --git a/python/paddle/incubate/distributed/fleet/base.py b/python/paddle/incubate/distributed/fleet/base.py
index 1bce9decb52908..87af4c7fd4e826 100644
--- a/python/paddle/incubate/distributed/fleet/base.py
+++ b/python/paddle/incubate/distributed/fleet/base.py
@@ -15,8 +15,8 @@
 import abc
 
 from paddle import base
-from paddle.distributed.fleet.base.role_maker import RoleMakerBase
 from paddle.base.executor import Executor
+from paddle.distributed.fleet.base.role_maker import RoleMakerBase
 from paddle.optimizer import SGD
 from paddle.static.amp.decorator import OptimizerWithMixedPrecision
 
diff --git a/python/paddle/incubate/distributed/fleet/collective.py b/python/paddle/incubate/distributed/fleet/collective.py
index 8ac688107e97b8..0a63ddb71dffb0 100644
--- a/python/paddle/incubate/distributed/fleet/collective.py
+++ b/python/paddle/incubate/distributed/fleet/collective.py
@@ -17,7 +17,6 @@
 import paddle
 import paddle.distributed.transpiler.distribute_transpiler as dist_transpiler
 from paddle import base
-from paddle.distributed.fleet.meta_optimizers import RawProgramOptimizer
 from paddle.base.compiler import CompiledProgram
 from paddle.base.executor import Executor
 from paddle.base.framework import Program
@@ -25,6 +24,7 @@
     CheckpointSaver,
     PaddleModel,
 )
+from paddle.distributed.fleet.meta_optimizers import RawProgramOptimizer
 from paddle.incubate.distributed.fleet.base import (
     DistributedOptimizer,
     Fleet,
diff --git a/python/paddle/incubate/distributed/fleet/fleet_util.py b/python/paddle/incubate/distributed/fleet/fleet_util.py
index 860c37705f4be8..687c1066f0c09a 100644
--- a/python/paddle/incubate/distributed/fleet/fleet_util.py
+++ b/python/paddle/incubate/distributed/fleet/fleet_util.py
@@ -25,8 +25,8 @@
 
 import paddle
 from paddle import base
-from paddle.distributed.fleet.utils.fs import HDFSClient
 from paddle.base.log_helper import get_logger
+from paddle.distributed.fleet.utils.fs import HDFSClient
 
 from . import utils
 
diff --git a/python/paddle/incubate/distributed/fleet/utils.py b/python/paddle/incubate/distributed/fleet/utils.py
index 4a60900f75435b..2be2ac7161071b 100644
--- a/python/paddle/incubate/distributed/fleet/utils.py
+++ b/python/paddle/incubate/distributed/fleet/utils.py
@@ -22,10 +22,10 @@
 
 import paddle
 from paddle import base
-from paddle.distributed.fleet.base.util_factory import draw_block_graphviz
 from paddle.base import core
 from paddle.base.framework import Program
 from paddle.base.proto import framework_pb2
+from paddle.distributed.fleet.base.util_factory import draw_block_graphviz
 from paddle.framework import io_utils
 
 __all__ = [
diff --git a/python/paddle/incubate/distributed/utils/io/dist_load.py b/python/paddle/incubate/distributed/utils/io/dist_load.py
index be37471ec0a600..4d1ce2a40ff6e4 100644
--- a/python/paddle/incubate/distributed/utils/io/dist_load.py
+++ b/python/paddle/incubate/distributed/utils/io/dist_load.py
@@ -17,8 +17,8 @@
 
 import paddle
 import paddle.distributed as dist
-from paddle.distributed import fleet
 from paddle.base.framework import dygraph_only
+from paddle.distributed import fleet
 
 
 @dygraph_only
diff --git a/python/paddle/incubate/distributed/utils/io/dist_save.py b/python/paddle/incubate/distributed/utils/io/dist_save.py
index f05828ea92bd09..94b07ed728cd26 100644
--- a/python/paddle/incubate/distributed/utils/io/dist_save.py
+++ b/python/paddle/incubate/distributed/utils/io/dist_save.py
@@ -18,9 +18,9 @@
 
 import paddle
 import paddle.distributed as dist
+from paddle.base.framework import dygraph_only
 from paddle.distributed import fleet
 from paddle.distributed.fleet.utils.log_util import logger
-from paddle.base.framework import dygraph_only
 
 from .save_for_auto import save_for_auto_inference
 
diff --git a/python/paddle/incubate/distributed/utils/io/save_for_auto.py b/python/paddle/incubate/distributed/utils/io/save_for_auto.py
index da127303ad1b78..5dd12d41218a6c 100644
--- a/python/paddle/incubate/distributed/utils/io/save_for_auto.py
+++ b/python/paddle/incubate/distributed/utils/io/save_for_auto.py
@@ -21,12 +21,12 @@
 
 import paddle
 import paddle.distributed as dist
+from paddle.base.framework import dygraph_only
 from paddle.distributed import fleet
 from paddle.distributed.fleet.meta_parallel.sharding.group_sharded_stage3 import (
     GroupShardedStage3,
 )
 from paddle.distributed.fleet.utils.log_util import logger
-from paddle.base.framework import dygraph_only
 
 __all__ = ["save_for_auto_inference"]
 
diff --git a/python/paddle/incubate/nn/functional/fused_dropout_add.py b/python/paddle/incubate/nn/functional/fused_dropout_add.py
index aa38a342937fe5..8546019a4959ed 100644
--- a/python/paddle/incubate/nn/functional/fused_dropout_add.py
+++ b/python/paddle/incubate/nn/functional/fused_dropout_add.py
@@ -14,8 +14,8 @@
 
 
 from paddle import _C_ops
-from paddle.common_ops_import import default_main_program
 from paddle.base import core
+from paddle.common_ops_import import default_main_program
 from paddle.framework import LayerHelper, in_dynamic_mode
 
 
diff --git a/python/paddle/jit/dy2static/convert_operators.py b/python/paddle/jit/dy2static/convert_operators.py
index 700c23189b0791..535c3713100e2c 100644
--- a/python/paddle/jit/dy2static/convert_operators.py
+++ b/python/paddle/jit/dy2static/convert_operators.py
@@ -16,10 +16,7 @@
 
 import paddle
 from paddle.base.data_feeder import convert_dtype
-from paddle.base.dygraph.base import (
-    _convert_into_variable,
-    in_declarative_mode,
-)
+from paddle.base.dygraph.base import _convert_into_variable, in_declarative_mode
 from paddle.base.framework import Variable, core, default_main_program
 
 from .utils import (
diff --git a/python/paddle/nn/clip.py b/python/paddle/nn/clip.py
index 2c27abbdc667f3..c00851e3019e67 100644
--- a/python/paddle/nn/clip.py
+++ b/python/paddle/nn/clip.py
@@ -19,9 +19,9 @@
 import paddle
 import paddle.autograd as imperative_base
 from paddle import _C_ops
-from paddle.common_ops_import import Variable, check_type, default_main_program
 from paddle.base import core, framework, unique_name
 from paddle.base.data_feeder import check_variable_and_dtype
+from paddle.common_ops_import import Variable, check_type, default_main_program
 from paddle.framework import LayerHelper, in_dynamic_mode
 from paddle.tensor.layer_function_generator import templatedoc
 
diff --git a/python/paddle/nn/functional/common.py b/python/paddle/nn/functional/common.py
index ca17169e23b7ad..3662fd919ba5a8 100644
--- a/python/paddle/nn/functional/common.py
+++ b/python/paddle/nn/functional/common.py
@@ -16,8 +16,8 @@
 
 import paddle
 from paddle import _C_ops
-from paddle.common_ops_import import Variable, default_main_program
 from paddle.base.layer_helper import LayerHelper
+from paddle.common_ops_import import Variable, default_main_program
 from paddle.framework import core, in_dynamic_mode
 from paddle.tensor.creation import full
 
diff --git a/python/paddle/nn/functional/conv.py b/python/paddle/nn/functional/conv.py
index f72b5d3822732b..77e58193a09bee 100644
--- a/python/paddle/nn/functional/conv.py
+++ b/python/paddle/nn/functional/conv.py
@@ -13,19 +13,19 @@
 # limitations under the License.
 
 from paddle import _C_ops, _legacy_C_ops, get_flags, in_dynamic_mode
+from paddle.base.framework import _global_flags
 from paddle.device import (
     get_all_custom_device_type,
     is_compiled_with_cuda,
     is_compiled_with_rocm,
 )
-from paddle.base.framework import _global_flags
 from paddle.tensor.manipulation import reshape
 from paddle.tensor.math import _add_with_axis
 
-from ...common_ops_import import Variable
-from ...device import get_cudnn_version
 from ...base.data_feeder import check_dtype, check_variable_and_dtype
 from ...base.layer_helper import LayerHelper
+from ...common_ops_import import Variable
+from ...device import get_cudnn_version
 from ...framework import no_grad
 from ...tensor.manipulation import squeeze, unsqueeze
 from ...utils import (
diff --git a/python/paddle/nn/functional/extension.py b/python/paddle/nn/functional/extension.py
index fbb6e4e412df37..de3d4dbe754e68 100644
--- a/python/paddle/nn/functional/extension.py
+++ b/python/paddle/nn/functional/extension.py
@@ -18,13 +18,13 @@
 
 from paddle import _C_ops, _legacy_C_ops, in_dynamic_mode
 
-from ...common_ops_import import Variable
 from ...base.data_feeder import (
     check_dtype,
     check_type,
     check_variable_and_dtype,
 )
 from ...base.layer_helper import LayerHelper
+from ...common_ops_import import Variable
 from ...framework import convert_np_dtype_to_dtype_, core
 from ...tensor.creation import assign
 
diff --git a/python/paddle/nn/functional/input.py b/python/paddle/nn/functional/input.py
index 57175ae79434d5..b76c2e5ea9b80f 100644
--- a/python/paddle/nn/functional/input.py
+++ b/python/paddle/nn/functional/input.py
@@ -14,9 +14,9 @@
 
 from paddle import _C_ops
 
-from ...common_ops_import import Variable
 from ...base.data_feeder import check_variable_and_dtype
 from ...base.layer_helper import LayerHelper
+from ...common_ops_import import Variable
 from ...framework import in_dynamic_mode
 
 __all__ = []
diff --git a/python/paddle/nn/functional/loss.py b/python/paddle/nn/functional/loss.py
index 4b572db25d79e6..2a0a2a64eec4a9 100644
--- a/python/paddle/nn/functional/loss.py
+++ b/python/paddle/nn/functional/loss.py
@@ -21,10 +21,10 @@
 from paddle.static.nn.control_flow import Assert
 from paddle.utils import deprecated
 
-from ...common_ops_import import Variable
 from ...base.data_feeder import check_variable_and_dtype
 from ...base.framework import _current_expected_place
 from ...base.layer_helper import LayerHelper
+from ...common_ops_import import Variable
 from ...tensor.manipulation import reshape
 
 __all__ = []
diff --git a/python/paddle/nn/functional/vision.py b/python/paddle/nn/functional/vision.py
index c382349d4622e9..d49ca156f77ea9 100644
--- a/python/paddle/nn/functional/vision.py
+++ b/python/paddle/nn/functional/vision.py
@@ -15,10 +15,10 @@
 from paddle import _C_ops, _legacy_C_ops, in_dynamic_mode
 from paddle.base.framework import in_dygraph_mode
 
-from ...common_ops_import import Variable
-from ...device import get_cudnn_version, is_compiled_with_rocm
 from ...base.data_feeder import check_variable_and_dtype
 from ...base.layer_helper import LayerHelper
+from ...common_ops_import import Variable
+from ...device import get_cudnn_version, is_compiled_with_rocm
 
 __all__ = []
 
diff --git a/python/paddle/nn/layer/layers.py b/python/paddle/nn/layer/layers.py
index 8908b38defec45..8e8c06ea2df497 100644
--- a/python/paddle/nn/layer/layers.py
+++ b/python/paddle/nn/layer/layers.py
@@ -2164,9 +2164,7 @@ def _transform(self, t, device, dtype, blocking):
 
         # 2. cast param / Tensor to dtype
         if dtype is not None and dtype != t_used.dtype:
-            with paddle.base.framework._dygraph_place_guard(
-                place=t_used.place
-            ):
+            with paddle.base.framework._dygraph_place_guard(place=t_used.place):
                 t_casted = t_used.cast(dtype=dtype)
         else:
             t_casted = t_used
diff --git a/python/paddle/nn/layer/rnn.py b/python/paddle/nn/layer/rnn.py
index df8800555992eb..3f5adf54d3968c 100644
--- a/python/paddle/nn/layer/rnn.py
+++ b/python/paddle/nn/layer/rnn.py
@@ -20,7 +20,6 @@
 
 import paddle
 from paddle import _C_ops, _legacy_C_ops, framework, in_dynamic_mode
-from paddle.common_ops_import import Variable
 from paddle.base.data_feeder import check_type, check_variable_and_dtype
 from paddle.base.dygraph.base import NON_PERSISTABLE_VAR_NAME_SUFFIX
 from paddle.base.framework import (
@@ -28,6 +27,7 @@
     in_dygraph_mode,
     program_guard,
 )
+from paddle.common_ops_import import Variable
 from paddle.framework import core
 from paddle.nn import functional as F
 from paddle.nn import initializer as I
diff --git a/python/paddle/signal.py b/python/paddle/signal.py
index 0dfc9262780082..b59f90915392ec 100644
--- a/python/paddle/signal.py
+++ b/python/paddle/signal.py
@@ -16,9 +16,9 @@
 from paddle import _C_ops
 from paddle.framework import in_dynamic_mode
 
-from .fft import fft_c2c, fft_c2r, fft_r2c
 from .base.data_feeder import check_variable_and_dtype
 from .base.layer_helper import LayerHelper
+from .fft import fft_c2c, fft_c2r, fft_r2c
 from .tensor.attribute import is_complex
 
 __all__ = [
diff --git a/python/paddle/sparse/unary.py b/python/paddle/sparse/unary.py
index 4fb2a8d32e1a0b..b97a588af1bbe4 100644
--- a/python/paddle/sparse/unary.py
+++ b/python/paddle/sparse/unary.py
@@ -16,13 +16,9 @@
 
 import paddle
 from paddle import _C_ops, in_dynamic_mode
-from paddle.common_ops_import import Variable
 from paddle.base.data_feeder import check_type, check_variable_and_dtype
-from paddle.base.framework import (
-    convert_np_dtype_to_dtype_,
-    core,
-    dygraph_only,
-)
+from paddle.base.framework import convert_np_dtype_to_dtype_, core, dygraph_only
+from paddle.common_ops_import import Variable
 from paddle.framework import LayerHelper
 
 __all__ = []
diff --git a/python/paddle/static/nn/common.py b/python/paddle/static/nn/common.py
index 0fff9d47277e11..e650c18f1ed292 100644
--- a/python/paddle/static/nn/common.py
+++ b/python/paddle/static/nn/common.py
@@ -19,11 +19,6 @@
 import numpy as np
 
 import paddle
-from paddle.common_ops_import import (
-    LayerHelper,
-    check_type,
-    check_variable_and_dtype,
-)
 from paddle.base import core, unique_name
 from paddle.base.data_feeder import check_dtype
 from paddle.base.framework import (
@@ -38,6 +33,11 @@
 from paddle.base.layers.layer_function_generator import templatedoc
 from paddle.base.param_attr import ParamAttr
 from paddle.base.wrapped_decorator import signature_safe_contextmanager
+from paddle.common_ops_import import (
+    LayerHelper,
+    check_type,
+    check_variable_and_dtype,
+)
 from paddle.nn.initializer import Constant, Normal
 
 __all__ = []
diff --git a/python/paddle/tensor/array.py b/python/paddle/tensor/array.py
index 752eb848aae743..f241071913341f 100644
--- a/python/paddle/tensor/array.py
+++ b/python/paddle/tensor/array.py
@@ -14,8 +14,8 @@
 
 # Define functions about array.
 
-from ..common_ops_import import Variable
 from ..base.data_feeder import check_type, check_variable_and_dtype
+from ..common_ops_import import Variable
 from ..framework import LayerHelper, core, in_dynamic_mode
 
 __all__ = []
diff --git a/python/paddle/tensor/attribute.py b/python/paddle/tensor/attribute.py
index 561681416f5c3c..60f142186809c0 100644
--- a/python/paddle/tensor/attribute.py
+++ b/python/paddle/tensor/attribute.py
@@ -19,9 +19,9 @@
 import paddle
 from paddle import _C_ops
 
-from ..common_ops_import import Variable
 from ..base.data_feeder import check_type, check_variable_and_dtype
 from ..base.framework import in_dygraph_mode
+from ..common_ops_import import Variable
 from ..framework import LayerHelper, core
 from .creation import _complex_to_real_dtype, assign
 
diff --git a/python/paddle/tensor/layer_function_generator.py b/python/paddle/tensor/layer_function_generator.py
index a1856cea7c4ede..02ab66eb1da2ad 100644
--- a/python/paddle/tensor/layer_function_generator.py
+++ b/python/paddle/tensor/layer_function_generator.py
@@ -18,9 +18,9 @@
 
 from paddle import _C_ops, _legacy_C_ops
 
-from ..common_ops_import import Variable
 from ..base.data_feeder import check_variable_and_dtype
 from ..base.proto import framework_pb2
+from ..common_ops_import import Variable
 from ..framework import (
     LayerHelper,
     OpProtoHolder,
diff --git a/python/paddle/tensor/linalg.py b/python/paddle/tensor/linalg.py
index 5bd05fd2b6ecf4..367072e3ed3705 100644
--- a/python/paddle/tensor/linalg.py
+++ b/python/paddle/tensor/linalg.py
@@ -18,12 +18,8 @@
 from paddle import _C_ops
 from paddle.common_ops_import import VarDesc
 
+from ..base.data_feeder import check_dtype, check_type, check_variable_and_dtype
 from ..common_ops_import import Variable
-from ..base.data_feeder import (
-    check_dtype,
-    check_type,
-    check_variable_and_dtype,
-)
 from ..framework import LayerHelper, in_dynamic_mode
 from .creation import full
 from .manipulation import cast
diff --git a/python/paddle/tensor/logic.py b/python/paddle/tensor/logic.py
index 153092193e1c1d..b0abc4fac23b4d 100755
--- a/python/paddle/tensor/logic.py
+++ b/python/paddle/tensor/logic.py
@@ -16,8 +16,8 @@
 
 import paddle
 
-from ..common_ops_import import Variable
 from ..base.data_feeder import check_type, check_variable_and_dtype
+from ..common_ops_import import Variable
 from .layer_function_generator import templatedoc
 
 Tensor = paddle.base.framework.core.eager.Tensor
diff --git a/python/paddle/tensor/math.py b/python/paddle/tensor/math.py
index f0cd9c031ac100..6296f102bff0b1 100644
--- a/python/paddle/tensor/math.py
+++ b/python/paddle/tensor/math.py
@@ -22,13 +22,13 @@
 from paddle.common_ops_import import VarDesc, dygraph_only, dygraph_utils
 from paddle.utils.inplace_utils import inplace_apis_in_dygraph_only
 
-from ..common_ops_import import Variable
 from ..base.data_feeder import (
     check_dtype,
     check_type,
     check_variable_and_dtype,
     convert_dtype,
 )
+from ..common_ops_import import Variable
 from ..framework import (
     LayerHelper,
     convert_np_dtype_to_dtype_,
diff --git a/python/paddle/tensor/random.py b/python/paddle/tensor/random.py
index e2b71457bbddee..bbfc533c5fc925 100644
--- a/python/paddle/tensor/random.py
+++ b/python/paddle/tensor/random.py
@@ -16,8 +16,8 @@
 
 import paddle
 from paddle import _C_ops, _legacy_C_ops
-from paddle.common_ops_import import Variable
 from paddle.base.framework import _current_expected_place
+from paddle.common_ops_import import Variable
 from paddle.framework import in_dynamic_mode
 
 from ..base.data_feeder import (
diff --git a/python/paddle/tensor/stat.py b/python/paddle/tensor/stat.py
index aa73e54c09e787..5b94f19c690e59 100644
--- a/python/paddle/tensor/stat.py
+++ b/python/paddle/tensor/stat.py
@@ -18,8 +18,8 @@
 from paddle import _C_ops, _ir_ops, ir
 from paddle.framework import in_dynamic_mode
 
-from ..common_ops_import import Variable
 from ..base.data_feeder import check_type, check_variable_and_dtype
+from ..common_ops_import import Variable
 from ..framework import LayerHelper, core
 from .math import _get_reduce_axis_with_tensor
 from .search import where
diff --git a/test/asp/test_fleet_with_asp_dynamic.py b/test/asp/test_fleet_with_asp_dynamic.py
index 6e46785ae2b3cd..03e8bbdcb8dd38 100644
--- a/test/asp/test_fleet_with_asp_dynamic.py
+++ b/test/asp/test_fleet_with_asp_dynamic.py
@@ -19,8 +19,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed import fleet
 from paddle.base import core
+from paddle.distributed import fleet
 from paddle.incubate.asp import ASPHelper
 
 cuda_visible_devices = os.getenv('CUDA_VISIBLE_DEVICES')
diff --git a/test/auto_parallel/test_dist_attr_v2.py b/test/auto_parallel/test_dist_attr_v2.py
index 7e0fc4a3be59e0..563c8a19019ceb 100644
--- a/test/auto_parallel/test_dist_attr_v2.py
+++ b/test/auto_parallel/test_dist_attr_v2.py
@@ -20,6 +20,7 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import nn, static
+from paddle.base.core import OperatorDistAttr, TensorDistAttr
 from paddle.distributed import fleet
 from paddle.distributed.auto_parallel.process_mesh import ProcessMesh
 from paddle.distributed.auto_parallel.static.dist_context import (
@@ -33,7 +34,6 @@
     _copy_dist_attr_to_cpp_for_graph,
 )
 from paddle.distributed.fleet import auto
-from paddle.base.core import OperatorDistAttr, TensorDistAttr
 
 paddle.enable_static()
 
diff --git a/test/auto_parallel/test_dist_matmul.py b/test/auto_parallel/test_dist_matmul.py
index 892ed45ce92a07..a60f942db104d4 100644
--- a/test/auto_parallel/test_dist_matmul.py
+++ b/test/auto_parallel/test_dist_matmul.py
@@ -15,9 +15,9 @@
 import unittest
 
 import paddle
-from paddle.distributed.fleet import auto
 from paddle.base import program_guard
 from paddle.base.backward import append_backward
+from paddle.distributed.fleet import auto
 
 paddle.enable_static()
 
diff --git a/test/auto_parallel/test_dist_op_cost.py b/test/auto_parallel/test_dist_op_cost.py
index b025eceb817c0e..6477d8646bca6c 100644
--- a/test/auto_parallel/test_dist_op_cost.py
+++ b/test/auto_parallel/test_dist_op_cost.py
@@ -16,14 +16,14 @@
 import unittest
 
 import paddle
+from paddle.base import program_guard
+from paddle.base.backward import append_backward
 from paddle.distributed.auto_parallel.static.cluster import Cluster
 from paddle.distributed.auto_parallel.static.operators.common import (
     get_distributed_operator_impl_container,
     is_elementwise_op,
 )
 from paddle.distributed.fleet import auto
-from paddle.base import program_guard
-from paddle.base.backward import append_backward
 
 paddle.enable_static()
 
diff --git a/test/auto_parallel/test_dist_pnorm.py b/test/auto_parallel/test_dist_pnorm.py
index eab9e7fabef0ad..004f05a56168bc 100644
--- a/test/auto_parallel/test_dist_pnorm.py
+++ b/test/auto_parallel/test_dist_pnorm.py
@@ -15,9 +15,9 @@
 import unittest
 
 import paddle
-from paddle.distributed.fleet import auto
 from paddle.base import program_guard
 from paddle.base.backward import append_backward
+from paddle.distributed.fleet import auto
 
 paddle.enable_static()
 
diff --git a/test/auto_parallel/test_pass_bf16.py b/test/auto_parallel/test_pass_bf16.py
index ce9841ced88c71..fd6f6daf41cd10 100644
--- a/test/auto_parallel/test_pass_bf16.py
+++ b/test/auto_parallel/test_pass_bf16.py
@@ -19,8 +19,8 @@
 
 import paddle
 from paddle import nn
-from paddle.distributed.fleet import auto
 from paddle.base import core
+from paddle.distributed.fleet import auto
 from paddle.static import InputSpec
 from paddle.static.amp.bf16.amp_utils import _valid_types
 from paddle.static.amp.fp16_utils import find_true_prev_op
diff --git a/test/auto_parallel/test_prim_dist_op.py b/test/auto_parallel/test_prim_dist_op.py
index a01e1fb90441fc..99e12b20998745 100644
--- a/test/auto_parallel/test_prim_dist_op.py
+++ b/test/auto_parallel/test_prim_dist_op.py
@@ -15,6 +15,7 @@
 import unittest
 
 import paddle
+from paddle.base.layer_helper import LayerHelper
 from paddle.distributed.auto_parallel.static.completion import Completer
 from paddle.distributed.auto_parallel.static.dist_context import (
     DistributedContext,
@@ -23,7 +24,6 @@
 from paddle.distributed.auto_parallel.static.partitioner import Partitioner
 from paddle.distributed.auto_parallel.static.utils import set_var_dist_attr
 from paddle.distributed.fleet import auto
-from paddle.base.layer_helper import LayerHelper
 from paddle.incubate.autograd import enable_prim
 
 paddle.enable_static()
diff --git a/test/auto_parallel/test_serialization.py b/test/auto_parallel/test_serialization.py
index 01c17b37335702..d246db2ed88840 100644
--- a/test/auto_parallel/test_serialization.py
+++ b/test/auto_parallel/test_serialization.py
@@ -19,6 +19,8 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import nn, static
+from paddle.base.core import TensorDistAttr
+from paddle.base.framework import Program
 from paddle.distributed import fleet
 from paddle.distributed.auto_parallel.static.dist_context import (
     DistributedContext,
@@ -26,8 +28,6 @@
 )
 from paddle.distributed.auto_parallel.static.process_mesh_v2 import ProcessMesh
 from paddle.distributed.fleet import auto
-from paddle.base.core import TensorDistAttr
-from paddle.base.framework import Program
 
 paddle.enable_static()
 
diff --git a/test/auto_parallel/test_shard_tensor_api.py b/test/auto_parallel/test_shard_tensor_api.py
index af96c715131c6e..86be83b41031ab 100644
--- a/test/auto_parallel/test_shard_tensor_api.py
+++ b/test/auto_parallel/test_shard_tensor_api.py
@@ -16,10 +16,10 @@
 
 import paddle
 import paddle.distributed as dist
+from paddle.base.dygraph.base import switch_to_static_graph
 from paddle.distributed.auto_parallel.static.dist_context import (
     get_default_distributed_context,
 )
-from paddle.fluid.dygraph.base import switch_to_static_graph
 
 
 class TestDistAttrBasic(unittest.TestCase):
diff --git a/test/collective/fleet/auto_parallel_parallelizer.py b/test/collective/fleet/auto_parallel_parallelizer.py
index a5bfe090d6904a..473da21fc52d74 100755
--- a/test/collective/fleet/auto_parallel_parallelizer.py
+++ b/test/collective/fleet/auto_parallel_parallelizer.py
@@ -17,9 +17,9 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import nn, static, utils
+from paddle.base import core
 from paddle.distributed import fleet
 from paddle.distributed.fleet import auto
-from paddle.base import core
 
 paddle.enable_static()
 _global_parallel_strategy = None
diff --git a/test/collective/fleet/dygraph_group_sharded_stage3.py b/test/collective/fleet/dygraph_group_sharded_stage3.py
index fb2c2fec6bd3b0..5587f51e1928b7 100644
--- a/test/collective/fleet/dygraph_group_sharded_stage3.py
+++ b/test/collective/fleet/dygraph_group_sharded_stage3.py
@@ -21,6 +21,7 @@
 import numpy as np
 
 import paddle
+from paddle.base import core
 from paddle.distributed.fleet.meta_parallel.sharding.group_sharded_optimizer_stage2 import (
     GroupShardedOptimizerStage2,
 )
@@ -33,7 +34,6 @@
 from paddle.distributed.fleet.meta_parallel.sharding.group_sharded_utils import (
     GroupShardedScaler,
 )
-from paddle.base import core
 from paddle.nn import Linear
 
 epoch = 10
diff --git a/test/collective/fleet/dygraph_group_sharded_stage3_offload.py b/test/collective/fleet/dygraph_group_sharded_stage3_offload.py
index b680f343e3f253..fc7791fd09aaa6 100644
--- a/test/collective/fleet/dygraph_group_sharded_stage3_offload.py
+++ b/test/collective/fleet/dygraph_group_sharded_stage3_offload.py
@@ -18,13 +18,13 @@
 import numpy as np
 
 import paddle
+from paddle.base import core
 from paddle.distributed.fleet.meta_parallel.sharding.group_sharded_stage3 import (
     GroupShardedStage3,
 )
 from paddle.distributed.fleet.meta_parallel.sharding.group_sharded_utils import (
     GroupShardedScaler,
 )
-from paddle.base import core
 from paddle.nn import Linear
 
 epoch = 10
diff --git a/test/collective/fleet/test_auto_checkpoint.py b/test/collective/fleet/test_auto_checkpoint.py
index ccb2bfb41a90db..5382e522612d7d 100644
--- a/test/collective/fleet/test_auto_checkpoint.py
+++ b/test/collective/fleet/test_auto_checkpoint.py
@@ -19,8 +19,8 @@
 
 import paddle
 import paddle.base.incubate.checkpoint.auto_checkpoint as acp
-from paddle.distributed.fleet.utils.fs import HDFSClient, LocalFS
 from paddle.base.incubate.checkpoint.checkpoint_saver import PaddleModel
+from paddle.distributed.fleet.utils.fs import HDFSClient, LocalFS
 
 paddle.enable_static()
 logger = get_logger()
diff --git a/test/collective/fleet/test_fleet_checkpoint.py b/test/collective/fleet/test_fleet_checkpoint.py
index 79c221fd152ac9..82f2f416e189cf 100644
--- a/test/collective/fleet/test_fleet_checkpoint.py
+++ b/test/collective/fleet/test_fleet_checkpoint.py
@@ -17,9 +17,9 @@
 
 import paddle
 from paddle import base
-from paddle.distributed.fleet.utils.fs import HDFSClient, LocalFS
 from paddle.base.incubate.checkpoint.auto_checkpoint import ExeTrainStatus
 from paddle.base.incubate.checkpoint.checkpoint_saver import CheckpointSaver
+from paddle.distributed.fleet.utils.fs import HDFSClient, LocalFS
 from paddle.incubate.distributed.fleet import role_maker
 from paddle.incubate.distributed.fleet.collective import fleet
 
diff --git a/test/collective/fleet/test_fleet_gradient_merge_meta_optimizer.py b/test/collective/fleet/test_fleet_gradient_merge_meta_optimizer.py
index 16e30f974e947f..fc8e116f1e9e09 100644
--- a/test/collective/fleet/test_fleet_gradient_merge_meta_optimizer.py
+++ b/test/collective/fleet/test_fleet_gradient_merge_meta_optimizer.py
@@ -75,9 +75,7 @@ def test_gm_pure_fp16_optimizer(self):
 
         params = train_prog.all_parameters()
         for param in train_prog.all_parameters():
-            self.assertEqual(
-                param.dtype, paddle.base.core.VarDesc.VarType.FP16
-            )
+            self.assertEqual(param.dtype, paddle.base.core.VarDesc.VarType.FP16)
 
         vars = [x.name for x in train_prog.list_vars()]
         self.assertIn('@GradientMerge', ''.join(vars))
diff --git a/test/collective/fleet/test_imperative_auto_mixed_precision_for_eager.py b/test/collective/fleet/test_imperative_auto_mixed_precision_for_eager.py
index 3d85c6cd01edd9..7bfdcf5723f2a6 100644
--- a/test/collective/fleet/test_imperative_auto_mixed_precision_for_eager.py
+++ b/test/collective/fleet/test_imperative_auto_mixed_precision_for_eager.py
@@ -1357,9 +1357,7 @@ def func_isinstance():
                     with paddle.amp.auto_cast(custom_white_list=['layer_norm']):
                         out = layer_norm(x)
 
-                    self.assertTrue(
-                        out.dtype == base.core.VarDesc.VarType.FP16
-                    )
+                    self.assertTrue(out.dtype == base.core.VarDesc.VarType.FP16)
 
         func_isinstance()
 
diff --git a/test/collective/process_group_mpi.py b/test/collective/process_group_mpi.py
index 5c0873c6234ce3..3f363e90df24b4 100644
--- a/test/collective/process_group_mpi.py
+++ b/test/collective/process_group_mpi.py
@@ -20,6 +20,8 @@
 
 import paddle
 import paddle.distributed as dist
+from paddle.base import core
+from paddle.base.framework import _set_expected_place
 from paddle.distributed.collective import (
     Group,
     _default_group_name,
@@ -27,8 +29,6 @@
     _set_group_map_backend,
     _set_group_map_by_name,
 )
-from paddle.base import core
-from paddle.base.framework import _set_expected_place
 
 ctypes.CDLL("libmpi.so", mode=ctypes.RTLD_GLOBAL)
 
diff --git a/test/distribution/test_distribution_bernoulli.py b/test/distribution/test_distribution_bernoulli.py
index 0419268674cca1..4845efcd6a29d0 100644
--- a/test/distribution/test_distribution_bernoulli.py
+++ b/test/distribution/test_distribution_bernoulli.py
@@ -27,9 +27,9 @@
 from test_distribution import DistributionNumpy
 
 import paddle
+from paddle.base.data_feeder import convert_dtype
 from paddle.distribution import Bernoulli
 from paddle.distribution.kl import kl_divergence
-from paddle.base.data_feeder import convert_dtype
 
 np.random.seed(2023)
 paddle.seed(2023)
diff --git a/test/distribution/test_distribution_cauchy.py b/test/distribution/test_distribution_cauchy.py
index a2ab7ca3d9b2d4..d0d570a37df079 100644
--- a/test/distribution/test_distribution_cauchy.py
+++ b/test/distribution/test_distribution_cauchy.py
@@ -27,9 +27,9 @@
 from test_distribution import DistributionNumpy
 
 import paddle
+from paddle.base.data_feeder import convert_dtype
 from paddle.distribution import Cauchy
 from paddle.distribution.kl import kl_divergence
-from paddle.base.data_feeder import convert_dtype
 
 np.random.seed(2023)
 paddle.seed(2023)
diff --git a/test/dygraph_to_static/test_bert.py b/test/dygraph_to_static/test_bert.py
index 2e0feffa24b583..c7b5272ff47659 100644
--- a/test/dygraph_to_static/test_bert.py
+++ b/test/dygraph_to_static/test_bert.py
@@ -28,9 +28,7 @@
 from paddle.base import core
 from paddle.jit.translated_layer import INFER_MODEL_SUFFIX, INFER_PARAMS_SUFFIX
 
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 SEED = 2020
 STEP_NUM = 10
 PRINT_STEP = 2
diff --git a/test/dygraph_to_static/test_cache_program.py b/test/dygraph_to_static/test_cache_program.py
index 0958cb0ff5716e..2b8a88245de87f 100644
--- a/test/dygraph_to_static/test_cache_program.py
+++ b/test/dygraph_to_static/test_cache_program.py
@@ -45,10 +45,7 @@ def test_cache(self):
                 # Check forward ops
                 prev_ops = cur_ops
                 cur_ops = Counter(
-                    [
-                        op.type
-                        for op in base.default_main_program().block(0).ops
-                    ]
+                    [op.type for op in base.default_main_program().block(0).ops]
                 )
                 if batch_id > 0:
                     prev_out_numpy = (
diff --git a/test/dygraph_to_static/test_cycle_gan.py b/test/dygraph_to_static/test_cycle_gan.py
index 9b8001d9f68b4a..3484b27d5fac5e 100644
--- a/test/dygraph_to_static/test_cycle_gan.py
+++ b/test/dygraph_to_static/test_cycle_gan.py
@@ -550,9 +550,7 @@ def optimizer_setting(parameters):
 
 def train(args, to_static):
     place = (
-        base.CUDAPlace(0)
-        if base.is_compiled_with_cuda()
-        else base.CPUPlace()
+        base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
     )
 
     paddle.jit.enable_to_static(to_static)
diff --git a/test/dygraph_to_static/test_dict.py b/test/dygraph_to_static/test_dict.py
index 576bbffda4c053..650d7c58b0ba19 100644
--- a/test/dygraph_to_static/test_dict.py
+++ b/test/dygraph_to_static/test_dict.py
@@ -20,9 +20,7 @@
 from paddle import base
 from paddle.jit import to_static
 
-PLACE = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+PLACE = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 
 
 class SubNetWithDict(paddle.nn.Layer):
diff --git a/test/dygraph_to_static/test_ifelse.py b/test/dygraph_to_static/test_ifelse.py
index 253cab46f6fccd..381ba5b8daef4c 100644
--- a/test/dygraph_to_static/test_ifelse.py
+++ b/test/dygraph_to_static/test_ifelse.py
@@ -19,6 +19,7 @@
 from ifelse_simple_func import (
     NetWithControlFlowIf,
     add_fn,
+    base,
     dyfunc_empty_nonlocal,
     dyfunc_ifelse_ret_int1,
     dyfunc_ifelse_ret_int2,
@@ -28,7 +29,6 @@
     dyfunc_with_if_else2,
     dyfunc_with_if_else3,
     dyfunc_with_if_else_with_list_generator,
-    base,
     if_tensor_case,
     if_with_and_or,
     if_with_and_or_1,
diff --git a/test/dygraph_to_static/test_mobile_net.py b/test/dygraph_to_static/test_mobile_net.py
index 0478be61f64281..607d7644763deb 100644
--- a/test/dygraph_to_static/test_mobile_net.py
+++ b/test/dygraph_to_static/test_mobile_net.py
@@ -496,9 +496,7 @@ class Args:
     print_step = 1
     train_step = 10
     place = (
-        base.CUDAPlace(0)
-        if base.is_compiled_with_cuda()
-        else base.CPUPlace()
+        base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
     )
     model_save_dir = None
     model_save_prefix = None
diff --git a/test/dygraph_to_static/test_resnet.py b/test/dygraph_to_static/test_resnet.py
index 4dc8170da9de98..3d4585117c9770 100644
--- a/test/dygraph_to_static/test_resnet.py
+++ b/test/dygraph_to_static/test_resnet.py
@@ -36,9 +36,7 @@
 # NOTE: Reduce batch_size from 8 to 2 to avoid unittest timeout.
 batch_size = 2
 epoch_num = 1
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 
 
 if base.is_compiled_with_cuda():
diff --git a/test/dygraph_to_static/test_resnet_amp.py b/test/dygraph_to_static/test_resnet_amp.py
index afb85d18b5921e..40c60520fbcadc 100644
--- a/test/dygraph_to_static/test_resnet_amp.py
+++ b/test/dygraph_to_static/test_resnet_amp.py
@@ -26,9 +26,7 @@
 # NOTE: Reduce batch_size from 8 to 2 to avoid unittest timeout.
 batch_size = 2
 epoch_num = 1
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 
 
 if base.is_compiled_with_cuda():
diff --git a/test/dygraph_to_static/test_save_inference_model.py b/test/dygraph_to_static/test_save_inference_model.py
index 23f8633a4e6810..8842b18b37a8f4 100644
--- a/test/dygraph_to_static/test_save_inference_model.py
+++ b/test/dygraph_to_static/test_save_inference_model.py
@@ -29,9 +29,7 @@
 
 np.random.seed(SEED)
 
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 
 
 class SimpleFcLayer(paddle.nn.Layer):
diff --git a/test/dygraph_to_static/test_save_load.py b/test/dygraph_to_static/test_save_load.py
index 7e1eae4858e835..c963b627d8b8fc 100644
--- a/test/dygraph_to_static/test_save_load.py
+++ b/test/dygraph_to_static/test_save_load.py
@@ -29,9 +29,7 @@
 
 np.random.seed(2020)
 
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 
 
 class PrimeNet(paddle.nn.Layer):
diff --git a/test/dygraph_to_static/test_se_resnet.py b/test/dygraph_to_static/test_se_resnet.py
index 80e7257962e62e..aef9b3a2f0b6a2 100644
--- a/test/dygraph_to_static/test_se_resnet.py
+++ b/test/dygraph_to_static/test_se_resnet.py
@@ -38,9 +38,7 @@
 PRINT_STEP = 2
 STEP_NUM = 10
 
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 
 # Note: Set True to eliminate randomness.
 #     1. For one operation, cuDNN has several algorithms,
diff --git a/test/dygraph_to_static/test_sentiment.py b/test/dygraph_to_static/test_sentiment.py
index 8e701e48b5b81c..f2692671729a8f 100644
--- a/test/dygraph_to_static/test_sentiment.py
+++ b/test/dygraph_to_static/test_sentiment.py
@@ -304,9 +304,7 @@ class Args:
 def train(args, to_static):
     paddle.jit.enable_to_static(to_static)
     place = (
-        base.CUDAPlace(0)
-        if base.is_compiled_with_cuda()
-        else base.CPUPlace()
+        base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
     )
 
     with base.dygraph.guard(place):
diff --git a/test/dygraph_to_static/test_seq2seq.py b/test/dygraph_to_static/test_seq2seq.py
index e39a4da16ba132..fee69b74bfdfec 100644
--- a/test/dygraph_to_static/test_seq2seq.py
+++ b/test/dygraph_to_static/test_seq2seq.py
@@ -25,9 +25,7 @@
 from paddle import base
 from paddle.nn import ClipGradByGlobalNorm
 
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 STEP_NUM = 10
 PRINT_STEP = 2
 
diff --git a/test/dygraph_to_static/test_transformer.py b/test/dygraph_to_static/test_transformer.py
index da5a3a7edacc64..af1918d982cc04 100644
--- a/test/dygraph_to_static/test_transformer.py
+++ b/test/dygraph_to_static/test_transformer.py
@@ -30,9 +30,7 @@
 from paddle import base
 
 trainer_count = 1
-place = (
-    base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
-)
+place = base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
 SEED = 10
 STEP_NUM = 10
 
diff --git a/test/dygraph_to_static/test_word2vec.py b/test/dygraph_to_static/test_word2vec.py
index 82fe2e1c0d516d..85edea2093d82f 100644
--- a/test/dygraph_to_static/test_word2vec.py
+++ b/test/dygraph_to_static/test_word2vec.py
@@ -281,9 +281,7 @@ def train(to_static):
     np.random.seed(0)
 
     place = (
-        base.CUDAPlace(0)
-        if base.is_compiled_with_cuda()
-        else base.CPUPlace()
+        base.CUDAPlace(0) if base.is_compiled_with_cuda() else base.CPUPlace()
     )
     with base.dygraph.guard(place):
         base.default_startup_program().random_seed = 1000
diff --git a/test/ir/inference/quant_dequant_test.py b/test/ir/inference/quant_dequant_test.py
index 2dfef926933ff2..4f1e2335f5d496 100644
--- a/test/ir/inference/quant_dequant_test.py
+++ b/test/ir/inference/quant_dequant_test.py
@@ -285,9 +285,7 @@ def check_output_with_option(
             paddle.reader.shuffle(paddle.dataset.mnist.train(), buf_size=500),
             batch_size=batch_size,
         )
-        feeder = base.DataFeeder(
-            feed_list=[self.data, self.label], place=place
-        )
+        feeder = base.DataFeeder(feed_list=[self.data, self.label], place=place)
         with base.scope_guard(scope):
             for _ in range(iters):
                 data = next(train_reader())
diff --git a/test/ir/inference/test_trt_c_allreduce_infer_script.py b/test/ir/inference/test_trt_c_allreduce_infer_script.py
index b82300ac9fb6a3..98b7e38e00c2c9 100644
--- a/test/ir/inference/test_trt_c_allreduce_infer_script.py
+++ b/test/ir/inference/test_trt_c_allreduce_infer_script.py
@@ -19,8 +19,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed import fleet
 from paddle.base import core
+from paddle.distributed import fleet
 from paddle.inference import Config, PrecisionType, create_predictor
 
 
diff --git a/test/ir/new_ir/test_pass_manager.py b/test/ir/new_ir/test_pass_manager.py
index 2f31e945f31f40..071afc5d7d6c98 100644
--- a/test/ir/new_ir/test_pass_manager.py
+++ b/test/ir/new_ir/test_pass_manager.py
@@ -16,7 +16,7 @@
 
 import paddle
 from paddle import ir
-from paddle.fluid import core
+from paddle.base import core
 from paddle.framework import LayerHelper
 
 paddle.enable_static()
diff --git a/test/legacy_test/detected_gpu.py b/test/legacy_test/detected_gpu.py
index a33916714ee45f..ff7d625805f6b1 100644
--- a/test/legacy_test/detected_gpu.py
+++ b/test/legacy_test/detected_gpu.py
@@ -19,10 +19,7 @@
 print("compile with cuda:", base.core.is_compiled_with_cuda())
 print("get_cuda_device_count:", base.core.get_cuda_device_count())
 
-if (
-    base.core.is_compiled_with_cuda()
-    and base.core.get_cuda_device_count() > 0
-):
+if base.core.is_compiled_with_cuda() and base.core.get_cuda_device_count() > 0:
     sys.exit(0)
 else:
     sys.exit(1)
diff --git a/test/legacy_test/dist_fleet_sync_batch_norm.py b/test/legacy_test/dist_fleet_sync_batch_norm.py
index 56a001178b4cfd..86e8e921adf10d 100644
--- a/test/legacy_test/dist_fleet_sync_batch_norm.py
+++ b/test/legacy_test/dist_fleet_sync_batch_norm.py
@@ -21,8 +21,8 @@
 
 import paddle
 from paddle import base
-from paddle.distributed import fleet
 from paddle.base import core
+from paddle.distributed import fleet
 from paddle.static import Executor, Program, program_guard
 
 
diff --git a/test/legacy_test/distributed_fused_lamb_test_base.py b/test/legacy_test/distributed_fused_lamb_test_base.py
index ec96fee570cf96..baffc7dd5e5460 100644
--- a/test/legacy_test/distributed_fused_lamb_test_base.py
+++ b/test/legacy_test/distributed_fused_lamb_test_base.py
@@ -19,9 +19,9 @@
 import numpy as np
 
 import paddle
+from paddle.base import core
 from paddle.distributed import fleet
 from paddle.distributed.fleet.meta_optimizers.common import CollectiveHelper
-from paddle.base import core
 from paddle.incubate import DistributedFusedLamb
 from paddle.nn.clip import ClipGradBase, _clip_by_global_norm_using_mp_type
 from paddle.vision.models import resnet18 as resnet
diff --git a/test/legacy_test/test_assign_pos_op.py b/test/legacy_test/test_assign_pos_op.py
index 57a4a75fb1bf3b..4b039eb1bd91d5 100644
--- a/test/legacy_test/test_assign_pos_op.py
+++ b/test/legacy_test/test_assign_pos_op.py
@@ -18,8 +18,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.models.moe import utils
 from paddle.base import core
+from paddle.distributed.models.moe import utils
 
 
 def assign_pos(x, _cum_count):
diff --git a/test/legacy_test/test_async_read_write.py b/test/legacy_test/test_async_read_write.py
index 39374fe1cdcf07..1af4e21c5c9a31 100644
--- a/test/legacy_test/test_async_read_write.py
+++ b/test/legacy_test/test_async_read_write.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.device import cuda
 from paddle.base import core
+from paddle.device import cuda
 
 
 class TestAsyncRead(unittest.TestCase):
diff --git a/test/legacy_test/test_auto_parallel_cost_model.py b/test/legacy_test/test_auto_parallel_cost_model.py
index d3b298309656be..8725abad91ba8d 100644
--- a/test/legacy_test/test_auto_parallel_cost_model.py
+++ b/test/legacy_test/test_auto_parallel_cost_model.py
@@ -17,6 +17,7 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import nn, static, utils
+from paddle.base import core
 from paddle.distributed import fleet
 from paddle.distributed.auto_parallel.static.completion import Completer
 from paddle.distributed.auto_parallel.static.cost_model import estimate_cost
@@ -29,7 +30,6 @@
 from paddle.distributed.auto_parallel.static.partitioner import Partitioner
 from paddle.distributed.auto_parallel.static.reshard import Resharder
 from paddle.distributed.fleet import auto
-from paddle.base import core
 
 paddle.enable_static()
 _global_parallel_strategy = "dp_mp_pp"
diff --git a/test/legacy_test/test_auto_parallel_mapper.py b/test/legacy_test/test_auto_parallel_mapper.py
index f80b637cc520cb..a1159f823f605e 100644
--- a/test/legacy_test/test_auto_parallel_mapper.py
+++ b/test/legacy_test/test_auto_parallel_mapper.py
@@ -22,6 +22,7 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import base, nn, static, utils
+from paddle.base import core
 from paddle.distributed import fleet
 from paddle.distributed.auto_parallel.static.cluster import Cluster
 from paddle.distributed.auto_parallel.static.completion import Completer
@@ -39,7 +40,6 @@
 from paddle.distributed.auto_parallel.static.partitioner import Partitioner
 from paddle.distributed.auto_parallel.static.reshard import Resharder
 from paddle.distributed.fleet import auto
-from paddle.base import core
 
 if os.getenv("CUDA_VISIBLE_DEVICES") is not None:
     os.environ["CUDA_VISIBLE_DEVICES"] = ""
diff --git a/test/legacy_test/test_auto_search_dist_op.py b/test/legacy_test/test_auto_search_dist_op.py
index 4567aafab59494..2fbf3c9d702938 100644
--- a/test/legacy_test/test_auto_search_dist_op.py
+++ b/test/legacy_test/test_auto_search_dist_op.py
@@ -16,6 +16,7 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import nn, static, utils
+from paddle.base import core
 from paddle.distributed.auto_parallel.static.dist_attribute import (
     OperatorDistAttr,
 )
@@ -23,7 +24,6 @@
 from paddle.distributed.auto_parallel.static.operators.common import (
     get_distributed_operator_impl_container,
 )
-from paddle.base import core
 
 paddle.enable_static()
 device = "gpu" if core.is_compiled_with_cuda() else "cpu"
diff --git a/test/legacy_test/test_base_layer.py b/test/legacy_test/test_base_layer.py
index 0ad517e00cdb1a..28b8d0cac762fd 100644
--- a/test/legacy_test/test_base_layer.py
+++ b/test/legacy_test/test_base_layer.py
@@ -392,9 +392,7 @@ def func_test_to_api(self):
             paddle.base.core.VarDesc.VarType.FP64,
         )
         for p in self.linear.parameters():
-            self.assertTrue(
-                isinstance(p, paddle.base.framework.EagerParamBase)
-            )
+            self.assertTrue(isinstance(p, paddle.base.framework.EagerParamBase))
 
         if paddle.base.is_compiled_with_cuda():
             self.linear.to(device=paddle.CUDAPlace(0))
@@ -470,9 +468,7 @@ def func_test_to_api_paddle_dtype(self):
             paddle.base.core.VarDesc.VarType.FP64,
         )
         for p in self.linear.parameters():
-            self.assertTrue(
-                isinstance(p, paddle.base.framework.EagerParamBase)
-            )
+            self.assertTrue(isinstance(p, paddle.base.framework.EagerParamBase))
 
     def func_test_to_api_numpy_dtype(self):
         self.linear.to(dtype=np.float64)
@@ -505,9 +501,7 @@ def func_test_to_api_numpy_dtype(self):
             paddle.base.core.VarDesc.VarType.FP64,
         )
         for p in self.linear.parameters():
-            self.assertTrue(
-                isinstance(p, paddle.base.framework.EagerParamBase)
-            )
+            self.assertTrue(isinstance(p, paddle.base.framework.EagerParamBase))
 
     def func_test_to_api_none_buffer(self):
         model = paddle.nn.Linear(2, 4)
diff --git a/test/legacy_test/test_boxps.py b/test/legacy_test/test_boxps.py
index 66582739c5dfa9..78c7473d4fcfd9 100644
--- a/test/legacy_test/test_boxps.py
+++ b/test/legacy_test/test_boxps.py
@@ -16,8 +16,8 @@
 
 import paddle
 from paddle import base
-from paddle.distributed.transpiler import collective
 from paddle.base import core
+from paddle.distributed.transpiler import collective
 from paddle.incubate.layers.nn import _pull_box_sparse
 
 
diff --git a/test/legacy_test/test_buffer_shared_memory_reuse_pass.py b/test/legacy_test/test_buffer_shared_memory_reuse_pass.py
index ab99363f751a04..4eaa5387216f0a 100644
--- a/test/legacy_test/test_buffer_shared_memory_reuse_pass.py
+++ b/test/legacy_test/test_buffer_shared_memory_reuse_pass.py
@@ -61,9 +61,7 @@ def build_program_and_scope(self):
 
                 with base.scope_guard(scope):
                     exe = base.Executor(
-                        base.CUDAPlace(0)
-                        if self.use_cuda
-                        else base.CPUPlace()
+                        base.CUDAPlace(0) if self.use_cuda else base.CPUPlace()
                     )
                     exe.run(startup_program)
 
diff --git a/test/legacy_test/test_checkpoint_saver.py b/test/legacy_test/test_checkpoint_saver.py
index 2fe34a3385822f..643ea78816579d 100644
--- a/test/legacy_test/test_checkpoint_saver.py
+++ b/test/legacy_test/test_checkpoint_saver.py
@@ -14,8 +14,8 @@
 
 import unittest
 
-from paddle.distributed.fleet.utils.fs import HDFSClient
 from paddle.base.incubate.checkpoint.checkpoint_saver import CheckpointSaver
+from paddle.distributed.fleet.utils.fs import HDFSClient
 
 
 class CheckpointerSaverTest(unittest.TestCase):
diff --git a/test/legacy_test/test_cost_model.py b/test/legacy_test/test_cost_model.py
index 1529d3432cd091..997a5c0c6c47b7 100644
--- a/test/legacy_test/test_cost_model.py
+++ b/test/legacy_test/test_cost_model.py
@@ -15,8 +15,8 @@
 import unittest
 
 import paddle
-from paddle.cost_model import CostModel
 from paddle.base import core
+from paddle.cost_model import CostModel
 
 paddle.enable_static()
 
diff --git a/test/legacy_test/test_cuda_graph_static_mode.py b/test/legacy_test/test_cuda_graph_static_mode.py
index 15df4acea3de91..746a3db02c2227 100644
--- a/test/legacy_test/test_cuda_graph_static_mode.py
+++ b/test/legacy_test/test_cuda_graph_static_mode.py
@@ -18,8 +18,8 @@
 from simple_nets import simple_fc_net_with_inputs
 
 import paddle
-from paddle.device.cuda.graphs import CUDAGraph
 from paddle.base.dygraph.base import switch_to_static_graph
+from paddle.device.cuda.graphs import CUDAGraph
 
 
 def can_use_cuda_graph():
diff --git a/test/legacy_test/test_cuda_graph_static_mode_error.py b/test/legacy_test/test_cuda_graph_static_mode_error.py
index f2ef98eab5f90a..a718f1b7009bdc 100644
--- a/test/legacy_test/test_cuda_graph_static_mode_error.py
+++ b/test/legacy_test/test_cuda_graph_static_mode_error.py
@@ -18,8 +18,8 @@
 from test_cuda_graph_static_mode import build_program, can_use_cuda_graph
 
 import paddle
-from paddle.device.cuda.graphs import CUDAGraph
 from paddle.base.dygraph.base import switch_to_static_graph
+from paddle.device.cuda.graphs import CUDAGraph
 
 
 class TestCUDAGraphInFirstBatch(unittest.TestCase):
diff --git a/test/legacy_test/test_cuda_max_memory_allocated.py b/test/legacy_test/test_cuda_max_memory_allocated.py
index b9f76325728b00..90e016921f8a21 100644
--- a/test/legacy_test/test_cuda_max_memory_allocated.py
+++ b/test/legacy_test/test_cuda_max_memory_allocated.py
@@ -15,12 +15,12 @@
 import unittest
 
 import paddle
+from paddle.base import core
 from paddle.device.cuda import (
     device_count,
     max_memory_allocated,
     memory_allocated,
 )
-from paddle.base import core
 
 
 class TestMaxMemoryAllocated(unittest.TestCase):
diff --git a/test/legacy_test/test_cuda_max_memory_reserved.py b/test/legacy_test/test_cuda_max_memory_reserved.py
index e6ccaddbfb985c..ac3b2b712e2ff7 100644
--- a/test/legacy_test/test_cuda_max_memory_reserved.py
+++ b/test/legacy_test/test_cuda_max_memory_reserved.py
@@ -15,12 +15,12 @@
 import unittest
 
 import paddle
+from paddle.base import core
 from paddle.device.cuda import (
     device_count,
     max_memory_reserved,
     memory_reserved,
 )
-from paddle.base import core
 
 
 class TestMaxMemoryreserved(unittest.TestCase):
diff --git a/test/legacy_test/test_cuda_memory_allocated.py b/test/legacy_test/test_cuda_memory_allocated.py
index bd5b294e735f87..3e4c2589406590 100644
--- a/test/legacy_test/test_cuda_memory_allocated.py
+++ b/test/legacy_test/test_cuda_memory_allocated.py
@@ -15,8 +15,8 @@
 import unittest
 
 import paddle
-from paddle.device.cuda import device_count, memory_allocated
 from paddle.base import core
+from paddle.device.cuda import device_count, memory_allocated
 
 
 class TestMemoryAllocated(unittest.TestCase):
diff --git a/test/legacy_test/test_cuda_memory_reserved.py b/test/legacy_test/test_cuda_memory_reserved.py
index 0a119aa3c30a0e..d639eab054ff52 100644
--- a/test/legacy_test/test_cuda_memory_reserved.py
+++ b/test/legacy_test/test_cuda_memory_reserved.py
@@ -15,8 +15,8 @@
 import unittest
 
 import paddle
-from paddle.device.cuda import device_count, memory_reserved
 from paddle.base import core
+from paddle.device.cuda import device_count, memory_reserved
 
 
 class TestMemoryreserved(unittest.TestCase):
diff --git a/test/legacy_test/test_dataset.py b/test/legacy_test/test_dataset.py
index b9a94767334e2a..fd4141c1c8b64d 100644
--- a/test/legacy_test/test_dataset.py
+++ b/test/legacy_test/test_dataset.py
@@ -285,9 +285,7 @@ def test_in_memory_dataset_run(self):
         else:
             for i in range(self.epoch_num):
                 try:
-                    exe.train_from_dataset(
-                        base.default_main_program(), dataset
-                    )
+                    exe.train_from_dataset(base.default_main_program(), dataset)
                 except Exception as e:
                     self.assertTrue(False)
 
@@ -552,9 +550,7 @@ def test_in_memory_dataset_run_2(self):
         else:
             for i in range(self.epoch_num):
                 try:
-                    exe.train_from_dataset(
-                        base.default_main_program(), dataset
-                    )
+                    exe.train_from_dataset(base.default_main_program(), dataset)
                 except Exception as e:
                     self.assertTrue(False)
 
@@ -638,9 +634,7 @@ def test_queue_dataset_run(self):
         else:
             for i in range(self.epoch_num):
                 try:
-                    exe.train_from_dataset(
-                        base.default_main_program(), dataset
-                    )
+                    exe.train_from_dataset(base.default_main_program(), dataset)
                 except Exception as e:
                     self.assertTrue(False)
 
@@ -710,9 +704,7 @@ def test_queue_dataset_run_2(self):
         else:
             for i in range(self.epoch_num):
                 try:
-                    exe.train_from_dataset(
-                        base.default_main_program(), dataset
-                    )
+                    exe.train_from_dataset(base.default_main_program(), dataset)
                 except Exception as e:
                     self.assertTrue(False)
 
@@ -776,9 +768,7 @@ def test_queue_dataset_run_3(self):
         else:
             for i in range(self.epoch_num):
                 try:
-                    exe.train_from_dataset(
-                        base.default_main_program(), dataset
-                    )
+                    exe.train_from_dataset(base.default_main_program(), dataset)
                 except Exception as e:
                     self.assertTrue(False)
 
diff --git a/test/legacy_test/test_debugger.py b/test/legacy_test/test_debugger.py
index bddd9ad540b1db..2acda1972fcc6f 100644
--- a/test/legacy_test/test_debugger.py
+++ b/test/legacy_test/test_debugger.py
@@ -14,9 +14,9 @@
 
 import unittest
 
-from paddle.distributed.fleet.base.util_factory import draw_block_graphviz
 from paddle.base import core
 from paddle.base.framework import Program
+from paddle.distributed.fleet.base.util_factory import draw_block_graphviz
 
 
 class TestDrawBlockGraphviz(unittest.TestCase):
diff --git a/test/legacy_test/test_dist_base.py b/test/legacy_test/test_dist_base.py
index 8e6a1c3c9c9ee9..2e6895b717579c 100755
--- a/test/legacy_test/test_dist_base.py
+++ b/test/legacy_test/test_dist_base.py
@@ -29,10 +29,10 @@
 
 import paddle
 from paddle import base
+from paddle.base import compiler
 from paddle.distributed.fleet.meta_optimizers import (
     RawProgramOptimizer as RawProgram,
 )
-from paddle.base import compiler
 from paddle.incubate.distributed.fleet import role_maker
 from paddle.incubate.distributed.fleet.collective import (
     DistributedStrategy,
diff --git a/test/legacy_test/test_einsum.py b/test/legacy_test/test_einsum.py
index 551c7948191f8f..830474de90f31d 100644
--- a/test/legacy_test/test_einsum.py
+++ b/test/legacy_test/test_einsum.py
@@ -179,9 +179,7 @@ def test_forward(self):
         expected_result = np.einsum(self.sample["paradigm"], *operands)
         equation = self.sample["paradigm"]
 
-        with paddle.base.dygraph.guard(
-            self._get_place(force_to_use_cpu=False)
-        ):
+        with paddle.base.dygraph.guard(self._get_place(force_to_use_cpu=False)):
             pd_operands = [paddle.to_tensor(operand) for operand in operands]
             result = paddle.einsum(equation, *pd_operands)
             self.check_output_equal(result.numpy(), expected_result)
@@ -348,9 +346,7 @@ def check_output_equal(self, actual, expect, rtol=1.0e-5, atol=1.0e-8):
 
     def check_output(self, eqn, *ops):
         expect = np.einsum(eqn, *ops)
-        with paddle.base.dygraph.guard(
-            self._get_place(force_to_use_cpu=False)
-        ):
+        with paddle.base.dygraph.guard(self._get_place(force_to_use_cpu=False)):
             pd_operands = [paddle.to_tensor(op) for op in ops]
             actual = paddle.einsum(eqn, *pd_operands)
             self.check_output_equal(actual.numpy(), expect)
diff --git a/test/legacy_test/test_einsum_v2.py b/test/legacy_test/test_einsum_v2.py
index cb93963b7dd81d..4ce8b7519bc8af 100644
--- a/test/legacy_test/test_einsum_v2.py
+++ b/test/legacy_test/test_einsum_v2.py
@@ -181,9 +181,7 @@ def test_forward(self):
         expected_result = np.einsum(self.sample["paradigm"], *operands)
         equation = self.sample["paradigm"]
 
-        with paddle.base.dygraph.guard(
-            self._get_place(force_to_use_cpu=False)
-        ):
+        with paddle.base.dygraph.guard(self._get_place(force_to_use_cpu=False)):
             pd_operands = [paddle.to_tensor(operand) for operand in operands]
             result = paddle.einsum(equation, *pd_operands)
             self.check_output_equal(result.numpy(), expected_result)
@@ -390,9 +388,7 @@ def check_output_equal(self, actual, expect, rtol=1.0e-5, atol=1.0e-8):
 
     def check_output(self, eqn, *ops):
         expect = np.einsum(eqn, *ops)
-        with paddle.base.dygraph.guard(
-            self._get_place(force_to_use_cpu=False)
-        ):
+        with paddle.base.dygraph.guard(self._get_place(force_to_use_cpu=False)):
             pd_operands = [paddle.to_tensor(op) for op in ops]
             actual = paddle.einsum(eqn, *pd_operands)
             self.check_output_equal(actual.numpy(), expect)
diff --git a/test/legacy_test/test_elementwise_add_op.py b/test/legacy_test/test_elementwise_add_op.py
index 4a9ecc51aa73aa..8630ef593be642 100644
--- a/test/legacy_test/test_elementwise_add_op.py
+++ b/test/legacy_test/test_elementwise_add_op.py
@@ -187,12 +187,8 @@ def setUp(self):
         self.axis = -1
 
         self.inputs = {
-            'X': OpTest.np_dtype_to_base_dtype(
-                convert_float_to_uint16(self.x)
-            ),
-            'Y': OpTest.np_dtype_to_base_dtype(
-                convert_float_to_uint16(self.y)
-            ),
+            'X': OpTest.np_dtype_to_base_dtype(convert_float_to_uint16(self.x)),
+            'Y': OpTest.np_dtype_to_base_dtype(convert_float_to_uint16(self.y)),
         }
         self.attrs = {'axis': self.axis, 'use_mkldnn': False}
         self.outputs = {'Out': convert_float_to_uint16(self.out)}
diff --git a/test/legacy_test/test_elementwise_mod_op.py b/test/legacy_test/test_elementwise_mod_op.py
index d75ae9e0a181ee..eb3ba740ea5e2c 100644
--- a/test/legacy_test/test_elementwise_mod_op.py
+++ b/test/legacy_test/test_elementwise_mod_op.py
@@ -177,12 +177,8 @@ def setUp(self):
         self.init_kernel_type()
         self.init_axis()
         self.inputs = {
-            'X': convert_float_to_uint16(
-                OpTest.np_dtype_to_base_dtype(self.x)
-            ),
-            'Y': convert_float_to_uint16(
-                OpTest.np_dtype_to_base_dtype(self.y)
-            ),
+            'X': convert_float_to_uint16(OpTest.np_dtype_to_base_dtype(self.x)),
+            'Y': convert_float_to_uint16(OpTest.np_dtype_to_base_dtype(self.y)),
         }
         self.attrs = {'axis': self.axis, 'use_mkldnn': self.use_mkldnn}
         self.outputs = {'Out': convert_float_to_uint16(self.out)}
diff --git a/test/legacy_test/test_elementwise_mul_op.py b/test/legacy_test/test_elementwise_mul_op.py
index 281087ce264204..6da50f02c163db 100644
--- a/test/legacy_test/test_elementwise_mul_op.py
+++ b/test/legacy_test/test_elementwise_mul_op.py
@@ -182,12 +182,8 @@ def setUp(self):
         self.axis = -1
 
         self.inputs = {
-            'X': OpTest.np_dtype_to_base_dtype(
-                convert_float_to_uint16(self.x)
-            ),
-            'Y': OpTest.np_dtype_to_base_dtype(
-                convert_float_to_uint16(self.y)
-            ),
+            'X': OpTest.np_dtype_to_base_dtype(convert_float_to_uint16(self.x)),
+            'Y': OpTest.np_dtype_to_base_dtype(convert_float_to_uint16(self.y)),
         }
         self.outputs = {'Out': convert_float_to_uint16(self.out)}
         self.attrs = {'axis': self.axis, 'use_mkldnn': False}
diff --git a/test/legacy_test/test_fetch_handler.py b/test/legacy_test/test_fetch_handler.py
index 5a41a6ea33b973..1d588a5d0339bf 100644
--- a/test/legacy_test/test_fetch_handler.py
+++ b/test/legacy_test/test_fetch_handler.py
@@ -52,9 +52,7 @@ def handler(self, fetch_dict):
             var_dict={'emb': var_emb, 'emb2': None, 'emb3': var_emb3},
             period_secs=1,
         )
-        default_fm = base.trainer_factory.FetchHandlerMonitor(
-            scope, default_fh
-        )
+        default_fm = base.trainer_factory.FetchHandlerMonitor(scope, default_fh)
         default_fm.start()
         time.sleep(5)
         default_fm.stop()
diff --git a/test/legacy_test/test_fleet_executor_cond_interceptor.py b/test/legacy_test/test_fleet_executor_cond_interceptor.py
index f6c02b5d815b02..463a7f29838796 100644
--- a/test/legacy_test/test_fleet_executor_cond_interceptor.py
+++ b/test/legacy_test/test_fleet_executor_cond_interceptor.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.fleet.fleet_executor_utils import TaskNode
 from paddle.base import core
+from paddle.distributed.fleet.fleet_executor_utils import TaskNode
 
 paddle.enable_static()
 
diff --git a/test/legacy_test/test_fleet_executor_task_node.py b/test/legacy_test/test_fleet_executor_task_node.py
index a15547f4f45516..1a2fa9be9bb461 100644
--- a/test/legacy_test/test_fleet_executor_task_node.py
+++ b/test/legacy_test/test_fleet_executor_task_node.py
@@ -15,8 +15,8 @@
 import unittest
 
 import paddle
-from paddle.distributed.fleet.fleet_executor_utils import TaskNode
 from paddle.base import core
+from paddle.distributed.fleet.fleet_executor_utils import TaskNode
 
 paddle.enable_static()
 
diff --git a/test/legacy_test/test_fused_attention_pass.py b/test/legacy_test/test_fused_attention_pass.py
index db938ab39974e6..37a356ea64b702 100644
--- a/test/legacy_test/test_fused_attention_pass.py
+++ b/test/legacy_test/test_fused_attention_pass.py
@@ -18,8 +18,8 @@
 
 import paddle
 import paddle.nn.functional as F
-from paddle.distributed.passes import PassManager, new_pass
 from paddle.base import core
+from paddle.distributed.passes import PassManager, new_pass
 
 paddle.enable_static()
 
diff --git a/test/legacy_test/test_fused_feedforward_pass.py b/test/legacy_test/test_fused_feedforward_pass.py
index 9fe700f6a683ed..03240d88d2d8be 100644
--- a/test/legacy_test/test_fused_feedforward_pass.py
+++ b/test/legacy_test/test_fused_feedforward_pass.py
@@ -18,8 +18,8 @@
 
 import paddle
 from paddle import nn
-from paddle.distributed.passes import PassManager, new_pass
 from paddle.base import core
+from paddle.distributed.passes import PassManager, new_pass
 
 paddle.enable_static()
 
diff --git a/test/legacy_test/test_get_device_properties.py b/test/legacy_test/test_get_device_properties.py
index 0cbfc4d16b0f44..41b7f94ad764c8 100644
--- a/test/legacy_test/test_get_device_properties.py
+++ b/test/legacy_test/test_get_device_properties.py
@@ -14,8 +14,8 @@
 
 import unittest
 
-from paddle.device.cuda import device_count, get_device_properties
 from paddle.base import core
+from paddle.device.cuda import device_count, get_device_properties
 
 
 class TestGetDeviceProperties(unittest.TestCase):
diff --git a/test/legacy_test/test_imperative_optimizer.py b/test/legacy_test/test_imperative_optimizer.py
index 8f6e37fef04d0a..a9af79fca6afac 100644
--- a/test/legacy_test/test_imperative_optimizer.py
+++ b/test/legacy_test/test_imperative_optimizer.py
@@ -19,8 +19,8 @@
 
 import paddle
 from paddle import base
-from paddle.distributed.fleet.meta_optimizers import DGCMomentumOptimizer
 from paddle.base import core
+from paddle.distributed.fleet.meta_optimizers import DGCMomentumOptimizer
 
 # Note(wangzhongpu)
 # In dygraph, don't support ModelAverage, DGCMomentumOptimizer, ExponentialMovingAverage, PipelineOptimizer, LookaheadOptimizer, RecomputeOptimizer.
diff --git a/test/legacy_test/test_imperative_optimizer_v2.py b/test/legacy_test/test_imperative_optimizer_v2.py
index 3262c5a05dd8ba..0000bd49ccb08d 100644
--- a/test/legacy_test/test_imperative_optimizer_v2.py
+++ b/test/legacy_test/test_imperative_optimizer_v2.py
@@ -20,8 +20,8 @@
 
 import paddle
 from paddle import base
-from paddle.distributed.fleet.meta_optimizers import DGCMomentumOptimizer
 from paddle.base import core
+from paddle.distributed.fleet.meta_optimizers import DGCMomentumOptimizer
 
 # Note(wangzhongpu)
 # In dygraph, don't support ModelAverage, DGCMomentumOptimizer, ExponentialMovingAverage, PipelineOptimizer, LookaheadOptimizer, RecomputeOptimizer.
diff --git a/test/legacy_test/test_inference_model_io.py b/test/legacy_test/test_inference_model_io.py
index 2f54e934818f36..2e179cf90276e9 100644
--- a/test/legacy_test/test_inference_model_io.py
+++ b/test/legacy_test/test_inference_model_io.py
@@ -22,13 +22,13 @@
 
 import paddle
 from paddle import base
+from paddle.base import core, executor
+from paddle.base.compiler import CompiledProgram
+from paddle.base.framework import Program, program_guard
 from paddle.distributed.io import (
     load_inference_model_distributed,
     save_persistables,
 )
-from paddle.base import core, executor
-from paddle.base.compiler import CompiledProgram
-from paddle.base.framework import Program, program_guard
 from paddle.static.io import load_inference_model, save_inference_model
 
 paddle.enable_static()
diff --git a/test/legacy_test/test_jit_save_load.py b/test/legacy_test/test_jit_save_load.py
index 1edb99e1ea77f4..e2df76f4751946 100644
--- a/test/legacy_test/test_jit_save_load.py
+++ b/test/legacy_test/test_jit_save_load.py
@@ -449,9 +449,7 @@ def tearDown(self):
         self.temp_dir.cleanup()
 
     def test_nest_output(self):
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
 
         net = LinearNetWithNestOut(8, 8)
         dy_outs = paddle.utils.flatten(net(x))
@@ -567,9 +565,7 @@ def test_with_input_spec(self):
 
         # 2. load to infer
         infer_layer = paddle.jit.load(model_path)
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         pred = infer_layer(x)
 
     def test_multi_in_out(self):
@@ -591,12 +587,8 @@ def test_multi_in_out(self):
 
         # 3. load to infer
         infer_layer = paddle.jit.load(model_path)
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
-        y = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
+        y = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         # 4. predict
         pred_x, pred_y = infer_layer(x, y)
 
@@ -633,12 +625,8 @@ def test_multi_in_out1(self):
 
         # 3. load to infer
         infer_layer = paddle.jit.load(model_path)
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
-        y = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
+        y = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         # 4. predict
         pred_x, pred_y = infer_layer(x, y)
 
@@ -680,9 +668,7 @@ def test_output_spec(self):
         adam = paddle.optimizer.Adam(
             learning_rate=0.1, parameters=train_layer.parameters()
         )
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         for i in range(10):
             out, loss = train_layer(x)
             loss.backward()
@@ -702,9 +688,7 @@ def test_output_spec(self):
 
         train_layer.eval()
         infer_layer = paddle.jit.load(model_path)
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         np.testing.assert_array_equal(
             train_layer(x)[0].numpy(), infer_layer(x).numpy()
         )
@@ -788,9 +772,7 @@ def train_and_save(self):
         adam = paddle.optimizer.Adam(
             learning_rate=0.1, parameters=train_layer.parameters()
         )
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         for i in range(10):
             hidden, loss = train_layer(x)
             loss.backward()
@@ -813,9 +795,7 @@ def test_load_pruned_model(self):
 
         infer_layer = paddle.jit.load(self.model_path)
 
-        x = base.dygraph.to_variable(
-            np.random.random((4, 8)).astype('float32')
-        )
+        x = base.dygraph.to_variable(np.random.random((4, 8)).astype('float32'))
         np.testing.assert_array_equal(
             train_layer(x)[0].numpy(), infer_layer(x).numpy()
         )
diff --git a/test/legacy_test/test_limit_by_capacity_op.py b/test/legacy_test/test_limit_by_capacity_op.py
index a897f35578a5b5..066345d5848246 100644
--- a/test/legacy_test/test_limit_by_capacity_op.py
+++ b/test/legacy_test/test_limit_by_capacity_op.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.models.moe import utils
 from paddle.base import core
+from paddle.distributed.models.moe import utils
 
 
 def limit_by_capacity(expert_count, _capacity, n_worker):
diff --git a/test/legacy_test/test_lookup_table_bf16_op.py b/test/legacy_test/test_lookup_table_bf16_op.py
index 204992f0c1c427..942a9cc3ccd74e 100644
--- a/test/legacy_test/test_lookup_table_bf16_op.py
+++ b/test/legacy_test/test_lookup_table_bf16_op.py
@@ -24,7 +24,7 @@
 from op import Operator
 
 import paddle
-from paddle import enable_static, base
+from paddle import base, enable_static
 from paddle.base import core
 
 
diff --git a/test/legacy_test/test_number_count_op.py b/test/legacy_test/test_number_count_op.py
index b877a40026dc2d..2ea68fd82d7ece 100644
--- a/test/legacy_test/test_number_count_op.py
+++ b/test/legacy_test/test_number_count_op.py
@@ -18,8 +18,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.models.moe import utils
 from paddle.base import core
+from paddle.distributed.models.moe import utils
 
 
 def count(x, upper_num):
diff --git a/test/legacy_test/test_paddle_save_load.py b/test/legacy_test/test_paddle_save_load.py
index f5d87ab70ea0f0..cb2e7e47afe864 100644
--- a/test/legacy_test/test_paddle_save_load.py
+++ b/test/legacy_test/test_paddle_save_load.py
@@ -570,9 +570,7 @@ def test_save_load_complex_object_dygraph_save(self):
             )
         self.assertTrue(load_tensor2['epoch'] == 123)
 
-        self.assertTrue(
-            isinstance(load_tensor3[0], paddle.base.core.LoDTensor)
-        )
+        self.assertTrue(isinstance(load_tensor3[0], paddle.base.core.LoDTensor))
         np.testing.assert_array_equal(
             np.array(load_tensor3[0]), obj3[0].numpy()
         )
diff --git a/test/legacy_test/test_parallel_executor_transformer.py b/test/legacy_test/test_parallel_executor_transformer.py
index cdf24938e25736..2cca066c2a9c70 100644
--- a/test/legacy_test/test_parallel_executor_transformer.py
+++ b/test/legacy_test/test_parallel_executor_transformer.py
@@ -21,8 +21,8 @@
 from parallel_executor_test_base import DeviceType, TestParallelExecutorBase
 
 import paddle
-from paddle.dataset import wmt16
 from paddle.base import core
+from paddle.dataset import wmt16
 
 os.environ['CPU_NUM'] = str(4)
 
diff --git a/test/legacy_test/test_prune_gate_by_capacity_op.py b/test/legacy_test/test_prune_gate_by_capacity_op.py
index 966651e22ff1c1..602c2ad787b602 100644
--- a/test/legacy_test/test_prune_gate_by_capacity_op.py
+++ b/test/legacy_test/test_prune_gate_by_capacity_op.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.models.moe import utils
 from paddle.base import core
+from paddle.distributed.models.moe import utils
 
 
 def count(x, upper_num):
diff --git a/test/legacy_test/test_py_func_op.py b/test/legacy_test/test_py_func_op.py
index 3164feab034b53..1706ad14d644d7 100644
--- a/test/legacy_test/test_py_func_op.py
+++ b/test/legacy_test/test_py_func_op.py
@@ -199,9 +199,7 @@ def test_main(use_cuda, use_py_func_op, use_parallel_executor):
             train_cp = base.default_main_program()
 
             if use_parallel_executor:
-                train_cp = compiler.CompiledProgram(
-                    base.default_main_program()
-                )
+                train_cp = compiler.CompiledProgram(base.default_main_program())
                 fetch_list = [loss.name]
             else:
                 fetch_list = [loss]
diff --git a/test/legacy_test/test_random_routing_op.py b/test/legacy_test/test_random_routing_op.py
index 5bc1752ecc9939..21a1746dd057f6 100644
--- a/test/legacy_test/test_random_routing_op.py
+++ b/test/legacy_test/test_random_routing_op.py
@@ -17,8 +17,8 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.models.moe import utils
 from paddle.base import core
+from paddle.distributed.models.moe import utils
 
 
 def random_routing(topk_idx, topk_value, prob, topk=2):
diff --git a/test/legacy_test/test_spawn_and_init_parallel_env.py b/test/legacy_test/test_spawn_and_init_parallel_env.py
index 8c9e3d4dde8362..69a35448b707d6 100644
--- a/test/legacy_test/test_spawn_and_init_parallel_env.py
+++ b/test/legacy_test/test_spawn_and_init_parallel_env.py
@@ -18,12 +18,12 @@
 
 import paddle
 import paddle.distributed as dist
+from paddle.base import core
 from paddle.distributed.spawn import (
     _get_default_nprocs,
     _get_subprocess_env_list,
     _options_valid_check,
 )
-from paddle.base import core
 
 # NOTE(chenweihang): Coverage CI is currently not able to count python3
 # unittest, so the unittests here covers some cases that will only be
diff --git a/test/legacy_test/test_sum_op.py b/test/legacy_test/test_sum_op.py
index db71738e61e001..fc374374090637 100644
--- a/test/legacy_test/test_sum_op.py
+++ b/test/legacy_test/test_sum_op.py
@@ -29,7 +29,7 @@
 
 import paddle
 import paddle.inference as paddle_infer
-from paddle import enable_static, base
+from paddle import base, enable_static
 from paddle.base import core
 from paddle.base.layer_helper import LayerHelper
 
diff --git a/test/legacy_test/test_where_op.py b/test/legacy_test/test_where_op.py
index 12fc6ea8878292..3daa2fbbb68982 100644
--- a/test/legacy_test/test_where_op.py
+++ b/test/legacy_test/test_where_op.py
@@ -148,9 +148,7 @@ def test_api(self, use_cuda=False):
                     result.stop_gradient = False
                     append_backward(paddle.mean(result))
                     for use_cuda in [False, True]:
-                        if use_cuda and (
-                            not base.core.is_compiled_with_cuda()
-                        ):
+                        if use_cuda and (not base.core.is_compiled_with_cuda()):
                             break
                         place = (
                             base.CUDAPlace(0) if use_cuda else base.CPUPlace()
diff --git a/test/mkldnn/test_layer_norm_bf16_mkldnn_op.py b/test/mkldnn/test_layer_norm_bf16_mkldnn_op.py
index f3bd7f731b3611..6e28472d5e9e03 100644
--- a/test/mkldnn/test_layer_norm_bf16_mkldnn_op.py
+++ b/test/mkldnn/test_layer_norm_bf16_mkldnn_op.py
@@ -24,7 +24,7 @@
     _reference_layer_norm_naive,
 )
 
-from paddle import enable_static, base
+from paddle import base, enable_static
 from paddle.base import core
 
 np.random.random(123)
diff --git a/test/mkldnn/test_layer_norm_mkldnn_op.py b/test/mkldnn/test_layer_norm_mkldnn_op.py
index 9db2b1966607d4..4533ccd05179d3 100644
--- a/test/mkldnn/test_layer_norm_mkldnn_op.py
+++ b/test/mkldnn/test_layer_norm_mkldnn_op.py
@@ -20,7 +20,7 @@
 import numpy as np
 from eager_op_test import OpTestTool, _set_use_system_allocator
 
-from paddle import enable_static, base
+from paddle import base, enable_static
 from paddle.base import core
 
 np.random.random(123)
diff --git a/test/prim/model/bert.py b/test/prim/model/bert.py
index 2b0805c6123e23..f7cf05f7ca2434 100644
--- a/test/prim/model/bert.py
+++ b/test/prim/model/bert.py
@@ -21,8 +21,8 @@
 import paddle
 import paddle.nn.functional as F
 from paddle import Tensor, nn
-from paddle.distributed.fleet.utils import recompute
 from paddle.base.data_feeder import convert_dtype
+from paddle.distributed.fleet.utils import recompute
 from paddle.io import DataLoader, Dataset
 from paddle.nn import MultiHeadAttention
 
diff --git a/test/prim/model/test_bert_cinn.py b/test/prim/model/test_bert_cinn.py
index c1bb108f58feec..3ae1bcb27aeea9 100644
--- a/test/prim/model/test_bert_cinn.py
+++ b/test/prim/model/test_bert_cinn.py
@@ -21,8 +21,8 @@
 
 import paddle
 from paddle import base
-from paddle.dataset.common import DATA_HOME, download
 from paddle.base import core
+from paddle.dataset.common import DATA_HOME, download
 
 SEED = 2023
 BATCH_SIZE = 2
diff --git a/test/prim/model/test_bert_prim.py b/test/prim/model/test_bert_prim.py
index 22e3cf856ae857..74a65e2f0761c8 100644
--- a/test/prim/model/test_bert_prim.py
+++ b/test/prim/model/test_bert_prim.py
@@ -21,8 +21,8 @@
 
 import paddle
 from paddle import base
-from paddle.dataset.common import DATA_HOME, download
 from paddle.base import core
+from paddle.dataset.common import DATA_HOME, download
 
 SEED = 2023
 BATCH_SIZE = 2
diff --git a/test/prim/model/test_bert_prim_cinn.py b/test/prim/model/test_bert_prim_cinn.py
index 21dd6c202baac1..46e8d63b7a5724 100644
--- a/test/prim/model/test_bert_prim_cinn.py
+++ b/test/prim/model/test_bert_prim_cinn.py
@@ -21,8 +21,8 @@
 
 import paddle
 from paddle import base
-from paddle.dataset.common import DATA_HOME, download
 from paddle.base import core
+from paddle.dataset.common import DATA_HOME, download
 
 SEED = 2023
 BATCH_SIZE = 2
diff --git a/test/quantization/quant2_int8_lstm_model.py b/test/quantization/quant2_int8_lstm_model.py
index 73659a8ac019cc..5bbb378e9c35e8 100644
--- a/test/quantization/quant2_int8_lstm_model.py
+++ b/test/quantization/quant2_int8_lstm_model.py
@@ -81,9 +81,7 @@ def get_warmup_tensor(self, data_path, place):
                     [len(feat) // 4 // 8, 8]
                 )
                 lod_feat = [feat.shape[0]]
-                minputs = paddle.base.create_lod_tensor(
-                    feat, [lod_feat], place
-                )
+                minputs = paddle.base.create_lod_tensor(feat, [lod_feat], place)
 
                 infer_data = core.PaddleTensor()
                 infer_data.lod = minputs.lod()
diff --git a/test/rnn/test_rnn_cells.py b/test/rnn/test_rnn_cells.py
index 4a0792b812d547..4bb6f49963f843 100644
--- a/test/rnn/test_rnn_cells.py
+++ b/test/rnn/test_rnn_cells.py
@@ -186,9 +186,7 @@ def runTest(self):
 
 def load_tests(loader, tests, pattern):
     suite = unittest.TestSuite()
-    devices = (
-        ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
-    )
+    devices = ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
     for bias in [True, False]:
         for device in devices:
             for test_class in [TestSimpleRNNCell, TestGRUCell, TestLSTMCell]:
diff --git a/test/rnn/test_rnn_cells_static.py b/test/rnn/test_rnn_cells_static.py
index 21a513e0c4e0bf..df0dbb11bbb511 100644
--- a/test/rnn/test_rnn_cells_static.py
+++ b/test/rnn/test_rnn_cells_static.py
@@ -342,9 +342,7 @@ def runTest(self):
 
 def load_tests(loader, tests, pattern):
     suite = unittest.TestSuite()
-    devices = (
-        ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
-    )
+    devices = ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
     for bias in [True, False]:
         for device in devices:
             for test_class in [TestSimpleRNNCell, TestGRUCell, TestLSTMCell]:
diff --git a/test/rnn/test_rnn_nets.py b/test/rnn/test_rnn_nets.py
index e75c83007b6a39..734dcae0fde56b 100644
--- a/test/rnn/test_rnn_nets.py
+++ b/test/rnn/test_rnn_nets.py
@@ -359,9 +359,7 @@ def forward(self, input):
 
 def load_tests(loader, tests, pattern):
     suite = unittest.TestSuite()
-    devices = (
-        ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
-    )
+    devices = ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
     for direction in ["forward", "bidirectional", "bidirect"]:
         for time_major in [True, False]:
             for device in devices:
diff --git a/test/rnn/test_rnn_nets_static.py b/test/rnn/test_rnn_nets_static.py
index b2b91685f3c5cb..20b8a7975e8c21 100644
--- a/test/rnn/test_rnn_nets_static.py
+++ b/test/rnn/test_rnn_nets_static.py
@@ -507,9 +507,7 @@ def runTest(self):
 
 def load_tests(loader, tests, pattern):
     suite = unittest.TestSuite()
-    devices = (
-        ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
-    )
+    devices = ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
     for direction in ["forward", "bidirectional", "bidirect"]:
         for time_major in [True, False]:
             for device in devices:
diff --git a/test/rnn/test_wrappers.py b/test/rnn/test_wrappers.py
index c6dd9692798deb..957d5894b98855 100644
--- a/test/rnn/test_wrappers.py
+++ b/test/rnn/test_wrappers.py
@@ -195,9 +195,7 @@ def runTest(self):
 
 def load_tests(loader, tests, pattern):
     suite = unittest.TestSuite()
-    devices = (
-        ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
-    )
+    devices = ["cpu", "gpu"] if paddle.base.is_compiled_with_cuda() else ["cpu"]
     for direction in ["forward", "backward"]:
         for device in devices:
             for time_major in [False]:
diff --git a/test/standalone_executor/test_standalone_executor_multi_micro_batch.py b/test/standalone_executor/test_standalone_executor_multi_micro_batch.py
index f814a18643ede2..f44e23ddf8dc24 100644
--- a/test/standalone_executor/test_standalone_executor_multi_micro_batch.py
+++ b/test/standalone_executor/test_standalone_executor_multi_micro_batch.py
@@ -19,10 +19,10 @@
 import numpy as np
 
 import paddle
-from paddle.distributed.passes.pass_utils import get_skip_gc_vars, split_program
 from paddle.base import core
 from paddle.base.core import Job, Plan
 from paddle.base.executor import _add_feed_fetch_ops, _StandaloneExecutor
+from paddle.distributed.passes.pass_utils import get_skip_gc_vars, split_program
 from paddle.nn import TransformerEncoderLayer
 
 paddle.enable_static()
diff --git a/tools/parse_kernel_info.py b/tools/parse_kernel_info.py
index 8a2edf06f7d894..23106ab0d2ebb0 100644
--- a/tools/parse_kernel_info.py
+++ b/tools/parse_kernel_info.py
@@ -94,9 +94,7 @@ def parse_paddle_kernels(lib="phi", kernel_type="function", print_detail=False):
     if lib == "phi":
         assert kernel_type in ["function", "structure", "all"]
         # phi kernel type can be: function, structure, all
-        kernel_infos = paddle.base.core._get_registered_phi_kernels(
-            kernel_type
-        )
+        kernel_infos = paddle.base.core._get_registered_phi_kernels(kernel_type)
     else:
         # fluid, phi, all
         assert kernel_type in ["fluid", "phi", "all"]