[CodeStyle][ruff] clean some F401 step: 9

pyproject.toml

@@ -122,25 +122,3 @@ known-first-party = ["paddle"]
 "test/dygraph_to_static/test_loop.py" = ["C416", "F821"]
 # Ignore unnecessary lambda in dy2st unittest test_lambda
 "test/dygraph_to_static/test_lambda.py" = ["PLC3002"]
-
-# temp ignore unused imports in all distributed files
-"python/paddle/distributed/transpiler/__init__.py" = ["F401"]
-"python/paddle/incubate/distributed/fleet/parameter_server/distribute_transpiler/__init__.py" = ["F401", "I001"]
-"python/paddle/distributed/fleet/runtime/__init__.py" = ["F401"]
-"python/paddle/distributed/transpiler/details/__init__.py" = ["F401", "I001"]
-"python/paddle/incubate/distributed/models/moe/gate/__init__.py" = ["F401", "I001"]
-"python/paddle/incubate/distributed/models/moe/__init__.py" = ["F401", "I001"]
-"python/paddle/incubate/distributed/utils/io/__init__.py" = ["F401", "I001"]
-"python/paddle/distributed/fleet/elastic/__init__.py" = ["F401", "I001"]
-
-# temp ignore isort
-"python/paddle/amp/__init__.py" = ["I001"]
-"python/paddle/distributed/__init__.py" = ["I001"]
-"python/paddle/distributed/communication/stream/__init__.py" = ["I001"]
-"python/paddle/distributed/launch/context/__init__.py" = ["I001"]
-"python/paddle/distributed/launch/controllers/__init__.py" = ["I001"]
-"python/paddle/distributed/passes/__init__.py" = ["I001"]
-"python/paddle/distributed/rpc/__init__.py" = ["I001"]
-"python/paddle/distribution/__init__.py" = ["I001"]
-"python/paddle/incubate/distributed/fleet/__init__.py" = ["I001"]
-"python/paddle/incubate/distributed/fleet/parameter_server/pslib/__init__.py" = ["I001"]

python/paddle/amp/__init__.py

@@ -12,35 +12,33 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from .auto_cast import (  # noqa: F401
-    auto_cast,
-    decorate,
-    amp_guard,
-    amp_decorate,
-)
-from .amp_lists import (  # noqa: F401
-    white_list,
-    black_list,
+from paddle.base import core
+from paddle.base.framework import (
+    _current_expected_place,
+    _get_paddle_place,
 )
 
 from . import (  # noqa: F401
+    accuracy_compare,
     debugging,
     grad_scaler,
-    accuracy_compare,
 )
-
+from .amp_lists import (  # noqa: F401
+    black_list,
+    white_list,
+)
+from .auto_cast import (  # noqa: F401
+    amp_decorate,
+    amp_guard,
+    auto_cast,
+    decorate,
+)
 from .grad_scaler import (  # noqa: F401
-    GradScaler,
     AmpScaler,
+    GradScaler,
     OptimizerState,
 )
 
-from paddle.base import core
-from paddle.base.framework import (
-    _current_expected_place,
-    _get_paddle_place,
-)
-
 __all__ = [
     'auto_cast',
     'GradScaler',

python/paddle/distributed/__init__.py

@@ -13,101 +13,93 @@
 # limitations under the License.
 
 import atexit  # noqa: F401
-from . import io
-from .spawn import spawn
-from .launch.main import launch
-from .parallel import (  # noqa: F401
-    init_parallel_env,
-    get_rank,
-    get_world_size,
-    ParallelEnv,
-    DataParallel,
-)
-from .parallel_with_gloo import (
-    gloo_init_parallel_env,
-    gloo_barrier,
-    gloo_release,
-)
 
-from paddle.distributed.fleet.dataset import InMemoryDataset, QueueDataset
-from paddle.distributed.fleet.base.topology import ParallelMode
+from paddle.base.core import Placement, ReduceType
 
+from . import (
+    cloud_utils,  # noqa: F401
+    io,
+    rpc,  # noqa: F401
+)
+from .auto_parallel import shard_op  # noqa: F401
+from .auto_parallel.api import (
+    DistAttr,
+    DistModel,
+    Strategy,
+    dtensor_from_fn,
+    reshard,
+    shard_layer,
+    shard_optimizer,
+    shard_tensor,
+    to_static,
+)
+from .auto_parallel.placement_type import (
+    Partial,
+    Replicate,
+    Shard,
+)
+from .auto_parallel.process_mesh import ProcessMesh
+from .checkpoint.load_state_dict import load_state_dict
+from .checkpoint.save_state_dict import save_state_dict
 from .collective import (
-    split,
-    new_group,
     is_available,
+    new_group,
+    split,
 )
-from .communication import (  # noqa: F401
-    stream,
+from .communication import (
+    P2POp,  # noqa: F401
     ReduceOp,
     all_gather,
     all_gather_object,
     all_reduce,
     alltoall,
     alltoall_single,
+    barrier,
+    batch_isend_irecv,  # noqa: F401
     broadcast,
     broadcast_object_list,
-    reduce,
-    send,
-    scatter,
+    destroy_process_group,
     gather,
-    scatter_object_list,
+    get_backend,
+    get_group,
+    irecv,
+    is_initialized,
     isend,
     recv,
-    irecv,
-    batch_isend_irecv,
-    P2POp,
+    reduce,
     reduce_scatter,
-    is_initialized,
-    destroy_process_group,
-    get_group,
+    scatter,
+    scatter_object_list,
+    send,
+    stream,  # noqa: F401
     wait,
-    barrier,
-    get_backend,
-)
-
-from .auto_parallel.process_mesh import ProcessMesh
-
-from paddle.base.core import ReduceType, Placement
-from .auto_parallel.placement_type import (
-    Shard,
-    Replicate,
-    Partial,
 )
-
-from .auto_parallel import shard_op  # noqa: F401
-
-from .auto_parallel.api import (
-    DistAttr,
-    shard_tensor,
-    dtensor_from_fn,
-    reshard,
-    shard_layer,
-    shard_optimizer,
-    to_static,
-    Strategy,
-    DistModel,
-)
-
-from .fleet import BoxPSDataset  # noqa: F401
-
-from .entry_attr import (  # noqa: F401
-    ProbabilityEntry,
+from .entry_attr import (
     CountFilterEntry,
+    ProbabilityEntry,
     ShowClickEntry,
 )
-
-from . import cloud_utils  # noqa: F401
-
+from .fleet import BoxPSDataset  # noqa: F401
+from .fleet.base.topology import ParallelMode
+from .fleet.dataset import InMemoryDataset, QueueDataset
+from .launch.main import launch
+from .parallel import (
+    DataParallel,  # noqa: F401
+    ParallelEnv,
+    get_rank,
+    get_world_size,
+    init_parallel_env,
+)
+from .parallel_with_gloo import (
+    gloo_barrier,
+    gloo_init_parallel_env,
+    gloo_release,
+)
 from .sharding import (  # noqa: F401
     group_sharded_parallel,
     save_group_sharded_model,
 )
-
-from . import rpc  # noqa: F401
-
-from .checkpoint.save_state_dict import save_state_dict
-from .checkpoint.load_state_dict import load_state_dict
+from .spawn import spawn
 
 __all__ = [
     "io",

python/paddle/distributed/auto_parallel/static/dist_context.py

@@ -15,9 +15,9 @@
 import copy
 from collections import defaultdict
 
-from paddle.distributed.passes import PassContext
 from paddle.framework import IrGraph, core, set_flags
 
+from ...passes import PassContext
 from ..process_mesh import ProcessMesh
 from .dist_op import DistributedOperator
 from .dist_tensor import DistributedTensor

python/paddle/distributed/communication/stream/__init__.py

@@ -16,12 +16,12 @@
 from .all_reduce import all_reduce
 from .all_to_all import alltoall, alltoall_single
 from .broadcast import broadcast
+from .gather import gather
+from .recv import recv
 from .reduce import reduce
 from .reduce_scatter import reduce_scatter
-from .recv import recv
 from .scatter import scatter
 from .send import send
-from .gather import gather
 
 __all__ = [
     "all_gather",

python/paddle/distributed/fleet/elastic/__init__.py

@@ -12,17 +12,19 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import signal
 import os
+import signal
 import sys
 
-from .manager import ElasticManager
-from .manager import ElasticStatus
-from .manager import ELASTIC_EXIT_CODE
-from .manager import ElasticLevel
-from .collective import CollectiveLauncher
+from paddle.distributed.fleet.launch_utils import DistributeMode  # noqa:F401
 
-from paddle.distributed.fleet.launch_utils import DistributeMode
+from .collective import CollectiveLauncher
+from .manager import (
+    ELASTIC_EXIT_CODE,
+    ElasticLevel,  # noqa:F401
+    ElasticManager,
+    ElasticStatus,
+)
 
 
 def enable_elastic(args, distribute_mode):

python/paddle/distributed/fleet/meta_optimizers/ps_optimizer.py

@@ -16,7 +16,7 @@
 import re
 import subprocess
 
-import paddle.distributed.passes
+import paddle
 from paddle.distributed.passes import PassContext
 from paddle.distributed.ps.utils.ps_factory import PsProgramBuilderFactory
 from paddle.distributed.ps.utils.public import (

python/paddle/distributed/fleet/runtime/__init__.py

@@ -12,8 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from .collective_runtime import CollectiveRuntime
-from .parameter_server_runtime import ParameterServerRuntime
-from .the_one_ps import TheOnePSRuntime
+from .collective_runtime import CollectiveRuntime  # noqa:F401
+from .parameter_server_runtime import ParameterServerRuntime  # noqa:F401
+from .the_one_ps import TheOnePSRuntime  # noqa:F401
 
 __all__ = []

python/paddle/distributed/launch/context/__init__.py

@@ -12,13 +12,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import logging
+
 from paddle.distributed.launch import plugins
 
+from .args_envs import env_args_mapping, fetch_envs, parse_args
 from .node import Node
 from .status import Status
-from .args_envs import parse_args, fetch_envs, env_args_mapping
-
-import logging
 
 
 class Context:

python/paddle/distributed/launch/controllers/__init__.py

@@ -14,10 +14,9 @@
 
 __all__ = []
 
-from .collective import CollectiveController
-from .collective import CollectiveElasticController
-from .ps import PSController
+from .collective import CollectiveController, CollectiveElasticController
 from .ipu_controller import IPUController
+from .ps import PSController
 from .rpc import RpcController
 
 # the order is extremely important

python/paddle/distributed/passes/__init__.py

@@ -12,27 +12,25 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from .pass_base import new_pass, PassManager, PassContext
-
-from .auto_parallel_gradient_merge import *  # noqa: F403
-from .auto_parallel_sharding import *  # noqa: F403
+from .allreduce_matmul_grad_overlapping import *  # noqa: F403
 from .auto_parallel_amp import *  # noqa: F403
-from .auto_parallel_fp16 import *  # noqa: F403
-from .auto_parallel_recompute import *  # noqa: F403
-from .auto_parallel_quantization import *  # noqa: F403
 from .auto_parallel_data_parallel_optimization import *  # noqa: F403
-from .auto_parallel_grad_clip import *  # noqa: F403
+from .auto_parallel_fp16 import *  # noqa: F403
 from .auto_parallel_fused_linear_promotion import *  # noqa: F403
-from .auto_parallel_supplement_explicit_dependencies import *  # noqa: F403
+from .auto_parallel_grad_clip import *  # noqa: F403
+from .auto_parallel_gradient_merge import *  # noqa: F403
 from .auto_parallel_pipeline import *  # noqa: F403
+from .auto_parallel_quantization import *  # noqa: F403
+from .auto_parallel_recompute import *  # noqa: F403
 from .auto_parallel_sequence_parallel_optimization import *  # noqa: F403
-from .allreduce_matmul_grad_overlapping import *  # noqa: F403
+from .auto_parallel_sharding import *  # noqa: F403
+from .auto_parallel_supplement_explicit_dependencies import *  # noqa: F403
 from .cpp_pass import *  # noqa: F403
 from .fuse_all_reduce import *  # noqa: F403
+from .pass_base import PassContext, PassManager, new_pass
 from .pipeline_scheduler_pass import *  # noqa: F403
-from .ps_trainer_pass import *  # noqa: F403
 from .ps_server_pass import *  # noqa: F403
-
+from .ps_trainer_pass import *  # noqa: F403
 
 __all__ = [
     'new_pass',