exchange trainA and trainB

[liu]

I want to exchange trainA and trainB, can we achieve the effect of darkening? Can the code be completed?
thanks!

[Yifan Jiang]

I dont try this before but you can definitely try and see hot it performs, by just exchanging the dataset

[liu]

I try it and fail. It dont work out very well

[Yifan Jiang]

maybe you could upload some results here

[liu]

loss does not converge，so....

[Yifan Jiang]

I think loss will never converge, since it is an adversarial training. Usually I only look at the visual results

[wzr1201]

hi,have you successfully achieved the effect of darkening?