I want to know exactly which transforms are applied to the validation set during training.
How do I know this?
For some transforms, it is explicitly mentioned how that transformation behaves in the validation set. For example, the OldRandomCrop
docs, it is explicitly mentioned that-
On the validation set, we take a center crop.
But this is not true for Zoom
transform. There’s no information on how it behaves on the validation set.
How do I get a list of what transformations are used in the validation set during the training, and equally importantly, how do I customize them and add transforms of my choice?