Xu Yang
Xu Yang
> In Stage 1, we trained on the non-accelerated model (sdxl-base) using traditional diffusion loss. For Stage 2 and 3, although in the first implementation all losses were calculated on...
@ToTheBeginning @guozinan126 @tsinggggg @zsxkib
besides, does the negative prompt participate in training
and stage1 model
where to get 'arcface-50' weights