Julian Fuchs
Results
1
comments of
Julian Fuchs
I think the paper is not very well explained in that point. As far as I understand it, the increase in dev loss is explained with overfitting the training data....