address precommit

Author: IwakuraRein

benchmarks/cutlass_benchmarks/w8a8_benchmarks.py

@@ -115,8 +115,10 @@ def bench_fp8(
     a_cont = a.contiguous()
     scale_a = torch.tensor(1.0, device="cuda", dtype=torch.float32)
     scale_b = torch.tensor(1.0, device="cuda", dtype=torch.float32)
+
     def ceil_div(x: int, y: int) -> int:
         return (x + y - 1) // y
+
     block_scale_a = torch.rand((m, ceil_div(k, 128)),
                                device="cuda",
                                dtype=torch.float32)

vllm/model_executor/layers/quantization/utils/fp8_utils.py

@@ -48,8 +48,10 @@ def apply_w8a8_block_fp8_linear(
     shape_supported_by_cutlass = (weight.shape[0] % 128 == 0
                                   and weight.shape[1] % 128 == 0)
     if current_platform.has_device_capability(100):
+
         def ceil_div(x: int, y: int) -> int:
             return (x + y - 1) // y
+
         shape_supported_by_cutlass = (
             ceil_div(weight.shape[0], 128) == weight_scale.shape[0]
             and ceil_div(weight.shape[1], 128) == weight_scale.shape[1])