[misc] update pre-commit and run all files (#4752)

* [misc] update pre-commit

* [misc] run pre-commit

* [misc] remove useless configuration files

* [misc] ignore cuda for clang-format

colossalai/fx/profiler/experimental/profiler_function/activation_function.py

@@ -1,5 +1,7 @@
 from typing import Tuple
+
 import torch
+
 from ..registry import meta_profiler_function
 
 # TODO: different activation has different FLOPs count, currently unused.

colossalai/fx/profiler/experimental/profiler_function/arithmetic.py

@@ -41,15 +41,15 @@ def _elementwise_flops_compute(input, other):
 @meta_profiler_function.register(torch.sub)
 @meta_profiler_function.register(torch.mul)
 @meta_profiler_function.register(torch.floor_divide)
-@meta_profiler_function.register('add')    # for built-in op +
-@meta_profiler_function.register('iadd')    # for built-in op +=
-@meta_profiler_function.register('eq')    # for built-in op =
-@meta_profiler_function.register('sub')    # for built-in op -
-@meta_profiler_function.register('isub')    # for built-in op -=
-@meta_profiler_function.register('mul')    # for built-in op *
-@meta_profiler_function.register('imul')    # for built-in op *=
-@meta_profiler_function.register('floordiv')    # for built-in op //
-@meta_profiler_function.register('ifloordiv')    # for built-in op //=
+@meta_profiler_function.register("add")  # for built-in op +
+@meta_profiler_function.register("iadd")  # for built-in op +=
+@meta_profiler_function.register("eq")  # for built-in op =
+@meta_profiler_function.register("sub")  # for built-in op -
+@meta_profiler_function.register("isub")  # for built-in op -=
+@meta_profiler_function.register("mul")  # for built-in op *
+@meta_profiler_function.register("imul")  # for built-in op *=
+@meta_profiler_function.register("floordiv")  # for built-in op //
+@meta_profiler_function.register("ifloordiv")  # for built-in op //=
 def torch_add_like_ops(input: Any, other: Any, *, out: Optional[torch.Tensor] = None) -> Tuple[int, int]:
     return _elementwise_flops_compute(input, other)
 
@@ -62,7 +62,7 @@ def torch_elementwise_op(input: torch.Tensor, *, out: Optional[torch.Tensor] = N
 
 
 @meta_profiler_function.register(torch.matmul)
-@meta_profiler_function.register('matmul')    # for built-in op @
+@meta_profiler_function.register("matmul")  # for built-in op @
 @meta_profiler_function.register(torch.Tensor.matmul)
 def torch_matmul(input: torch.Tensor, other: torch.Tensor, *, out: Optional[torch.Tensor] = None) -> Tuple[int, int]:
     macs = reduce(operator.mul, input.shape) * other.shape[-1]
@@ -78,13 +78,15 @@ def torch_bmm(input: torch.Tensor, other: torch.Tensor, *, out: Optional[torch.T
 
 
 @meta_profiler_function.register(torch.var_mean)
-def torch_var_mean(input: torch.Tensor,
-                   dim: Union[int, Tuple[int, ...]],
-                   unbiased: Optional[bool] = True,
-                   keepdim: Optional[bool] = False,
-                   *,
-                   out: Optional[torch.Tensor] = None) -> Tuple[int, int]:
-    assert out is None, 'saving to out is not supported yet'
+def torch_var_mean(
+    input: torch.Tensor,
+    dim: Union[int, Tuple[int, ...]],
+    unbiased: Optional[bool] = True,
+    keepdim: Optional[bool] = False,
+    *,
+    out: Optional[torch.Tensor] = None,
+) -> Tuple[int, int]:
+    assert out is None, "saving to out is not supported yet"
     flops = input.numel() * 3
     macs = 0
     return flops, macs

colossalai/fx/profiler/experimental/profiler_function/embedding.py

@@ -1,5 +1,7 @@
-import torch
 from typing import Optional
+
+import torch
+
 from ..registry import meta_profiler_function
 
 

colossalai/fx/profiler/experimental/profiler_function/linear.py

@@ -1,5 +1,7 @@
 from typing import Tuple
+
 import torch
+
 from ..registry import meta_profiler_function
 
 

colossalai/fx/profiler/experimental/profiler_function/normalization.py

@@ -1,5 +1,7 @@
 from typing import List, Optional, Tuple
+
 import torch
+
 from ..registry import meta_profiler_function
 
 
@@ -21,11 +23,13 @@ def torch_nn_func_instancenorm(
 
 
 @meta_profiler_function.register(torch.nn.functional.group_norm)
-def torch_nn_func_groupnorm(input: torch.Tensor,
-                            num_groups: int,
-                            weight: Optional[torch.Tensor] = None,
-                            bias: Optional[torch.Tensor] = None,
-                            eps: float = 1e-5) -> Tuple[int, int]:
+def torch_nn_func_groupnorm(
+    input: torch.Tensor,
+    num_groups: int,
+    weight: Optional[torch.Tensor] = None,
+    bias: Optional[torch.Tensor] = None,
+    eps: float = 1e-5,
+) -> Tuple[int, int]:
     has_affine = weight is not None
     flops = input.numel() * (5 if has_affine else 4)
     macs = 0

colossalai/fx/profiler/experimental/profiler_function/pooling.py

@@ -1,5 +1,7 @@
-from typing import Tuple, Union
+from typing import Tuple
+
 import torch
+
 from ..registry import meta_profiler_function
 
 

colossalai/fx/profiler/experimental/profiler_function/python_ops.py

@@ -1,6 +1,6 @@
 import operator
 from typing import Any, Tuple
-import torch
+
 from ..registry import meta_profiler_function
 
 

colossalai/fx/profiler/experimental/profiler_function/torch_ops.py

@@ -1,7 +1,9 @@
-from functools import reduce
 import operator
+from functools import reduce
 from typing import Any, Optional, Tuple
+
 import torch
+
 from ..registry import meta_profiler_function
 
 
@@ -43,13 +45,11 @@ def torch_where(condition: torch.Tensor, x: Any, y: Any) -> Tuple[int, int]:
 
 
 @meta_profiler_function.register(torch.max)
-def torch_max(input: torch.Tensor,
-              dim: int = None,
-              keepdim: bool = False,
-              *,
-              out: Optional[torch.Tensor] = None) -> Tuple[int, int]:
+def torch_max(
+    input: torch.Tensor, dim: int = None, keepdim: bool = False, *, out: Optional[torch.Tensor] = None
+) -> Tuple[int, int]:
     macs = 0
-    assert out is None, 'assigning value to out is not supported yet'
+    assert out is None, "assigning value to out is not supported yet"
     if dim is not None:
         shape = list(input.shape)
         shape.pop(int(dim))