np.float32 required?

[JoeThanks67]

data = np.asarray(data, dtype=np.float32) # {-1, 1}

Is it possible to use np.int8 or something else here? The range of values is only -1/1.