I have a NN here:
model = keras.models.Sequential([ keras.layers.Conv2D(64, 7, activation="relu", padding="same",