How to handle failure mode

[solomatov]

What's the best practice for handling failure mode, i.e. when the discriminator or the generator has 0 loss. Should we just terminate learning and debug the algorithm? Or should we rollback a couple of epochs back and continue training from there?

[John1231983]

@solomatov : Do you find the solution? I have same issue as you

[solomatov]

@John1231983 No, eventually I fixed the algorithm, but I don't remember what I did there.

[solomatov]

@John1231983 If you are interested, it's here: https://github.com/solomatov/cs231n-project