Do we need to train VQGAN and CLIP together?
charlotte12l opened this issue · 2 comments
charlotte12l commented
In training, do we need to train VQGAN and CLIP together? Or they are just trained separately but when inference we will use them two.
nerdyrodent commented
They're pre-trained.
charlotte12l commented
@nerdyrodent Actually I mean when I finetune(train) the model, do I just train VQGAN and CLIP separately, or train them together? I guess just separately is enough?