About training

[Carringbrink]

Hello author, thank you very much for your work on DIS, I have a questions about the training of this paper that I would like to ask you:

After I overfitted ISNetGTEncoder, I was going to train ISNetDIS, but I found that the loss for  fs and dfs feature layer is very large, what was your initial loss at that time?

[xuebinqin]

please keep training after 2000 iterations, you can check the mae and Fmeasure. If they are in the reasonable range, e.g. F\belong[0.4,0.8], There should be an issue about the loss.

…

On Sun, Jul 24, 2022 at 8:12 PM Carringbrink ***@***.***> wrote: Hello author, thank you very much for your work on DIS, I have a questions about the training of this paper that I would like to ask you: After I overfitted ISNetGTEncoder, I was going to train ISNetDIS, but I found that the loss for fs and dfs feature layer is very large, what was your initial loss at that time? [image: image] <https://user-images.githubusercontent.com/104354080/180691783-5f734f49-26d6-45c3-8794-6f89b8f3bb81.png> [image: image] <https://user-images.githubusercontent.com/104354080/180691749-8617b224-bf60-4692-8b27-efb0082dc8c8.png> — Reply to this email directly, view it on GitHub <#15>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADSGORO6BZS2SSL7MEBP4N3VVYAZ3ANCNFSM54QW4GIQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Xuebin Qin PhD Department of Computing Science University of Alberta, Edmonton, AB, Canada Homepage: https://xuebinqin.github.io/