active-learning icon indicating copy to clipboard operation
active-learning copied to clipboard

Published 20 hours ago • verified publisher icon google

Fixes random seeding

Open jarnoux opened this issue 2 years ago • 0 comments

self.already_selected is initialized to [] in the constructor but is never assigned None so that will always be false.

If already_selected is empty, ind will be initialized to np.argmax(self.min_distances), which was initialized to None and never reassigned so the first selected datapoint will always be np.argmax(None) = 0

jarnoux avatar Dec 05 '22 01:12 jarnoux