I noticed that "no grad" is used throughout the script. The newer "inference mode" is better if you're just running inference. Is it possible to use this or would it interfere too much with some of the scripts that are used for training purposes? Possible to use both depending on running inference versus training? Thanks.
I noticed that "no grad" is used throughout the script. The newer "inference mode" is better if you're just running inference. Is it possible to use this or would it interfere too much with some of the scripts that are used for training purposes? Possible to use both depending on running inference versus training? Thanks.
For your reference:
https://pytorch.org/docs/stable/generated/torch.autograd.grad_mode.inference_mode.html