How to prepare data and perform training of autoencoder?

Question

I have a dataset of images that looks like this:

array([[[[0.35980392, 0.26078431, 0.14313725],
     [0.38137255, 0.26470588, 0.15196078],
     [0.51960784, 0