fix: rm `use_cuda` param #2678

Job	Run time
pytorch_tutorial_build_worker (1, 15, linux.g5.12xlarge.nvidia.gpu)	15m 55s
pytorch_tutorial_build_worker (2, 15, linux.g5.4xlarge.nvidia.gpu)	50m 41s
pytorch_tutorial_build_worker (3, 15, linux.g5.4xlarge.nvidia.gpu)	19m 40s
pytorch_tutorial_build_worker (4, 15, linux.g5.4xlarge.nvidia.gpu)	27m 51s
pytorch_tutorial_build_worker (5, 15, linux.g5.4xlarge.nvidia.gpu)	18m 15s
pytorch_tutorial_build_worker (6, 15, linux.g5.4xlarge.nvidia.gpu)	17m 18s
pytorch_tutorial_build_worker (7, 15, linux.4xlarge.nvidia.gpu)	20m 9s
pytorch_tutorial_build_worker (8, 15, linux.4xlarge.nvidia.gpu)	16m 42s
pytorch_tutorial_build_worker (9, 15, linux.4xlarge.nvidia.gpu)	35m 42s
pytorch_tutorial_build_worker (10, 15, linux.4xlarge.nvidia.gpu)	37m 34s
pytorch_tutorial_build_worker (11, 15, linux.4xlarge.nvidia.gpu)	28m 24s
pytorch_tutorial_build_worker (12, 15, linux.4xlarge.nvidia.gpu)	18m 11s
pytorch_tutorial_build_worker (13, 15, linux.4xlarge.nvidia.gpu)	22m 49s
pytorch_tutorial_build_worker (14, 15, linux.4xlarge.nvidia.gpu)	13m 49s
pytorch_tutorial_build_worker (15, 15, linux.4xlarge.nvidia.gpu)	21m 0s
pytorch_tutorial_build_manager	19m 52s
	6h 23m 52s

Provide feedback