[NFC] polish code format

[NFC] polish code format

colossalai/cli/cli.py

 import click
-from .launcher import run
-from .check import check
+
 from .benchmark import benchmark
+from .check import check
+from .launcher import run
 
 
 class Arguments():

colossalai/cli/launcher/__init__.py

 import click
-from .run import launch_multi_processes
+
 from colossalai.context import Config
 
+from .run import launch_multi_processes
+
 
 @click.command(help="Launch distributed training on a single node or multiple nodes",
                context_settings=dict(ignore_unknown_options=True))

colossalai/context/moe_context.py

+from typing import Tuple
+
 import torch
 import torch.distributed as dist
 
@@ -5,8 +7,6 @@ from colossalai.context.parallel_mode import ParallelMode
 from colossalai.context.singleton_meta import SingletonMeta
 from colossalai.tensor import ProcessGroup
 
-from typing import Tuple
-
 
 def _check_sanity():
     from colossalai.core import global_context as gpc

colossalai/context/process_group_initializer/initializer_2d.py

@@ -2,10 +2,11 @@ import math
 
 import torch.distributed as dist
 
+from colossalai.global_variables import tensor_parallel_env as env
 from colossalai.registry import DIST_GROUP_INITIALIZER
-from .process_group_initializer import ProcessGroupInitializer
+
 from ..parallel_mode import ParallelMode
-from colossalai.global_variables import tensor_parallel_env as env
+from .process_group_initializer import ProcessGroupInitializer
 
 
 def _check_summa_env_var(summa_dim):

colossalai/context/process_group_initializer/initializer_pipeline.py

@@ -4,8 +4,9 @@
 from torch import distributed as dist
 
 from colossalai.registry import DIST_GROUP_INITIALIZER
-from .process_group_initializer import ProcessGroupInitializer
+
 from ..parallel_mode import ParallelMode
+from .process_group_initializer import ProcessGroupInitializer
 
 
 @DIST_GROUP_INITIALIZER.register_module

colossalai/context/process_group_initializer/initializer_sequence.py

@@ -3,9 +3,10 @@
 import torch.distributed as dist
 
 from colossalai.registry import DIST_GROUP_INITIALIZER
+
+from ..parallel_mode import ParallelMode
 from .initializer_tensor import Initializer_Tensor
 from .process_group_initializer import ProcessGroupInitializer
-from ..parallel_mode import ParallelMode
 
 
 @DIST_GROUP_INITIALIZER.register_module

colossalai/engine/gradient_handler/utils.py

+from typing import Iterable
+
 import torch.distributed as dist
 import torch.nn as nn
 from torch._utils import _flatten_dense_tensors, _unflatten_dense_tensors
-from typing import Iterable
 
 
 def bucket_allreduce(param_list: Iterable[nn.Parameter], group=None):