Contents:
Get torch computation device automatically based on GPU availability and memory usage
n (int) – Number of GPUs to request
int
tuple[str, list[int] | str]
tuple
str
list
Accelerator type
List of granted devices