Hello,
at 6:39 of the chapter Improved implementation of softmax, prof Andrew said tensorflow can avoid very large or small numbers.
I can’t understand this thing.
Tensorflow is supposed to give a more accurate value.Then how does avoiding work?
1 Like