Can I do with torch.no_grad():
optimizer.zero_grad()
loss.backward()
optimizer.step()
Can I do with torch.no_grad():
optimizer.zero_grad()
loss.backward()
optimizer.step()
No that won’t work, as this disables gradient calculation.
From the docs:
Context-manager that disabled gradient calculation.
Disabling gradient calculation is useful for inference, when you are sure that you will not callTensor.backward()
. It will reduce memory consumption for computations that would otherwise have requires_grad=True. In this mode, the result of every computation will have requires_grad=False, even when the inputs have requires_grad=True.
What is your intention?
I was just wanted to implement dataloader on the fly but I found a better solution without much headache. It’s a DataLoader from PyTorch that handles all the problems of creating target and input by providing a folder name with folders as categories.
Yeah I know the one you are talking about - DL4J also has a similar dataloading capability if you ever need it.