I have an image set of 256 that gets augmented to a size of 512.
Then in a custom loss function.
Images, list = batch
Where images is of size (512,3,96,96) (batch size, layers, h, w)
When i perform convolution and normalization i get [512,128] for a size().
I would like to split the standard pictures from the augmented ones to perform a custom loss function and was wondering how i would go about doing this.
If it was better to split the tensor or try and split images in the list… i tried split() but when i did i split it into [512,64] and i don’t know if this is the right way of splitting them.