where to max the PriorDiscriminator?

[bojone]

the prior term of deep infomax is a minimax game, like GAN. But I can not find any code about it in train.py. Do I have a wrong understand or your implement has some problem?

[mm04926412]

Equally confused, this error has propogated to some "child" projects if it is indeed an error.

[rcalland]

you're right, I made a mistake and there should be an update step for the prior discriminator