albertz
commented
11 months ago I always preferred import numpy over import numpy as np, as it is more readable to me. It's also more consistent to import torch (but inconsistent to import tensorflow as tf, although you could argue that tensorflow is longer than torch, which has the same length as numpy).
(This is just my opinion. I see that the other argument of following more the official recommendation also makes a lot of sense. I don't really have such a strong opinion on this.)
See https://github.com/rwth-i6/i6_models/pull/37#discussion_r1356588208
We should have the numpy import consistent, and I would go with the established default of
import numpy as np, although some of our other repos might have that still different.