JunlinHan

If you have a 24GB memory GPU, that's possible. or you can load your dataset in 800x600 resolution, then crop images to 600x450, 16GB should be fine.

> custom just use batch_size 1, do not try large batch_size at first. As usual, split two domains to trainA and trainB.

This is due to your environment. I would suggest you try setting num_threads to 0 and reinstall some packages (follow the requirements).

Hi! DCLGAN is quite memory sensitive, it needs to support two generators/discriminators at a single GPU. If you have a GPU with 16G memory, you can support the training in...

> > > Hi, > why do bigger patches needed for 512*512 res images? > what is the best iteration per image ratio what I mean is how many epochs...

> > > Hi，thank you for your excellent work！ > > If the width and height of the image I want are different, how should I train? Do I need...

> > > Hi, thanks for your reply on the front of crop size and load size. How does the network react in changing generator power? > when we face...

> > > In respect of generator arch, you mean using more blocks (such as 12 - 18 blocks) or even more filters(NGF = 80 or 128) in Resnet doesn't...

Hi kk2487, Thanks for your questions. Parameters of G_A and G_B are chained together in the optimizer. loss_G calculates both loss of G_A and G_B. Thus they can be backpropagate...

> Should G_A backpropagate with loss_G_A and G_B backpropagate with loss_G_B and update the paramerters separately? Yes, the parameters should be updated separately. Here the implementation is actually identical. (...