I think that fine-tuning the whole model (a.k.a. Dreambooth) on iPhone would require more RAM / processing power than it currently has. More viable path is to i

I think that fine-tuning the whole model (a.k.a. Dreambooth) on iPhone would require more RAM / processing power than it currently has. More viable path is to implement Hypernetwork + Textual Inversion, that is within possibility of today's hardware.