alias gpu to CUDAAccelerator

Author: justusschock

src/pytorch_lightning/accelerators/cuda.py

@@ -97,6 +97,12 @@ def register_accelerators(cls, accelerator_registry: Dict) -> None:
             cls,
             description=f"{cls.__class__.__name__}",
         )
+        # temporarily enable "gpu" to point to the CUDA Accelerator
+        accelerator_registry.register(
+            "gpu",
+            cls,
+            description=f"{cls.__class__.__name__}",
+        )
 
     def teardown(self) -> None:
         # clean up memory