About 0.01 uses per million words.
The learned weights of a neural network are often considered devoid of scrutable internal structure.
decipherable graspable comprehendible understandable comprehensible comprehendable tangible intelligible interpretable understandible
and east