def content_loss(Y_hat, Y):
    return np.square(Y_hat - Y).mean()