Does PyTorch provide any tutorials on using CUDA?
I haven’t been able to find any.
I have a serial PyTorch neural net code that I’d like to parallelize with CUDA. Would this generally be straightforward to do?
I don’t know what kind of tutorial you want exactly, but PyTorch Tutorial provides basic information on how to run training and validation on GPU.
To be short, all you need is
device = torch.device('cuda:0')
model.to(device)
# on training or validation epoch
inputs, labels = inputs.to(device), labels.to(device)
Hope this answered your question.
2 Likes