kacper

Results 1 issues of kacper

You mentioned in the paper, that you sample minibatch size N_s = 128. I was wondering whether whole batch contributes to the regression loss in your implementation or samples marked...