[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/kernel/jit/__init__.py

@@ -1,8 +1,10 @@
-from .option import set_jit_fusion_options
-from .bias_dropout_add import bias_dropout_add_fused_train, bias_dropout_add_fused_inference
+from .bias_dropout_add import bias_dropout_add_fused_inference, bias_dropout_add_fused_train
 from .bias_gelu import bias_gelu_impl
+from .option import set_jit_fusion_options
 
 __all__ = [
-    "bias_dropout_add_fused_train", "bias_dropout_add_fused_inference", "bias_gelu_impl",
-    "set_jit_fusion_options"
+    "bias_dropout_add_fused_train",
+    "bias_dropout_add_fused_inference",
+    "bias_gelu_impl",
+    "set_jit_fusion_options",
 ]

colossalai/kernel/jit/bias_dropout_add.py

@@ -1,5 +1,4 @@
 import torch
-from torch import Tensor
 
 
 def bias_dropout_add(x, bias, residual, prob, training):
@@ -10,16 +9,14 @@ def bias_dropout_add(x, bias, residual, prob, training):
 
 
 @torch.jit.script
-def bias_dropout_add_fused_train(x: torch.Tensor,
-                                 bias: torch.Tensor,
-                                 residual: torch.Tensor,
-                                 prob: float) -> torch.Tensor:
+def bias_dropout_add_fused_train(
+    x: torch.Tensor, bias: torch.Tensor, residual: torch.Tensor, prob: float
+) -> torch.Tensor:
     return bias_dropout_add(x, bias, residual, prob, True)
 
 
 @torch.jit.script
-def bias_dropout_add_fused_inference(x: torch.Tensor,
-                                     bias: torch.Tensor,
-                                     residual: torch.Tensor,
-                                     prob: float) -> torch.Tensor:
+def bias_dropout_add_fused_inference(
+    x: torch.Tensor, bias: torch.Tensor, residual: torch.Tensor, prob: float
+) -> torch.Tensor:
     return bias_dropout_add(x, bias, residual, prob, False)

colossalai/kernel/jit/bias_gelu.py

@@ -29,7 +29,6 @@ def bias_gelu_back(g, bias, y):
 
 
 class GeLUFunction(torch.autograd.Function):
-
     @staticmethod
     # bias is an optional argument
     def forward(ctx, input, bias):

colossalai/kernel/jit/option.py

@@ -10,15 +10,14 @@ JIT_OPTIONS_SET = False
 
 
 def set_jit_fusion_options():
-    """Set PyTorch JIT layer fusion options.
-    """
+    """Set PyTorch JIT layer fusion options."""
     # LSG: the latest pytorch and CUDA versions may not support
     # the following jit settings
     global JIT_OPTIONS_SET
     if JIT_OPTIONS_SET == False:
         # flags required to enable jit fusion kernels
-        TORCH_MAJOR = int(torch.__version__.split('.')[0])
-        TORCH_MINOR = int(torch.__version__.split('.')[1])
+        TORCH_MAJOR = int(torch.__version__.split(".")[0])
+        TORCH_MINOR = int(torch.__version__.split(".")[1])
         if (TORCH_MAJOR > 1) or (TORCH_MAJOR == 1 and TORCH_MINOR >= 10):
             # nvfuser
             torch._C._jit_set_profiling_executor(True)
@@ -38,12 +37,14 @@ def set_jit_fusion_options():
         JIT_OPTIONS_SET = True
 
 
-def warmup_jit_fusion(batch_size: int,
-                      hidden_size: int,
-                      seq_length: int = 512,
-                      vocab_size: int = 32768,
-                      dtype: torch.dtype = torch.float32):
-    """ Compile JIT functions before the main training steps """
+def warmup_jit_fusion(
+    batch_size: int,
+    hidden_size: int,
+    seq_length: int = 512,
+    vocab_size: int = 32768,
+    dtype: torch.dtype = torch.float32,
+):
+    """Compile JIT functions before the main training steps"""
 
     embed = Embedding(vocab_size, hidden_size).to(get_current_device())
     linear_1 = Linear(hidden_size, hidden_size * 4, skip_bias_add=True).to(get_current_device())