image = load_image("cat.png") # e.g. a 224x224x3 grid of numbers
 normalized = image / 255.0 # scale pixel values to [0, 1]

 flattened = flatten(normalized) # or keep as a grid for a convolutional l
 ayer
