import torch
print(torch.cuda.is_available())  # True if CUDA is available
print(torch.cuda.get_device_name(0))  # GPU name
print(torch.version.cuda)  # CUDA version PyTorch was compiled with