i\'m trying to train a dense network for images.
the train set shape returns:
train_X.shape (73257, 32, 32)
and the network architectu