I'm practicing to use CNN in image recognition.
no_training_model = keras.Sequential()
no_training_model.add(keras.layers.Conv2D(32, kernel_size=3, activation='relu', padding='same', input_shape=(28,28,1)
no_training_conv = no_training_model.layers[0]
no_training_weights = no_training_conv.weights[0].numpy()
in 4th row, what's the difference when I use to_numpy() instead?
What does .numpy() do anyway?