I'm attempting to build a CNN deep learning model utilizing a genetic algorithm to learn on CIFAR-10. I have so far found a few libraries, but I have mainly encountered pyGAD and some derivatives built on pyGAD. In the examples of pyGAD implementations I could find, the training step always utilizes a tensor object.
I know that a dataset item normally needs to be iterated, as answered in question 74258668, but pyGAD does not appear to support processing such a method from what I could find.
Should I be using a different approach for this? Am I importing/processing the dataset wrong? Should I switch the genetic algorithm source from pyGAD to something else?
Note: I am using torchvision to import the CIFAR-10 dataset, and I have already converted the individual items from the dataloader version to tensor objects. I was previously following the CIFAR-10 pytorch tutorial.