I was trying to recreate figure.3 in the paper and I found that simply changing the random seed would influence the plot of a same modle. And I noticed that the images was taken from trainloader, which includes data argumentation, so I would like to know that whether you use it in fig.3? Thanks!
I was trying to recreate figure.3 in the paper and I found that simply changing the random seed would influence the plot of a same modle. And I noticed that the images was taken from trainloader, which includes data argumentation, so I would like to know that whether you use it in fig.3? Thanks!