Oh right, I forgot some diffusion models can't offload / split layers. I don't use vision generation models much at all - was just going off LLM work. Apologies

Oh right, I forgot some diffusion models can't offload / split layers. I don't use vision generation models much at all - was just going off LLM work. Apologies for the potential misinformation.