diff --git a/cutedsl/bridge.py b/cutedsl/bridge.py index 1cab54d1..f8807c8f 100644 --- a/cutedsl/bridge.py +++ b/cutedsl/bridge.py @@ -21,7 +21,6 @@ import threading import torch import cutlass import cutlass.cute as cute -import cutlass.cute.backend # noqa: F401 (triggers CUDA init) import cutlass_torch from cutedsl.kernel.moe.torch_scaled_grouped_mm import (