This model was trained using the diffusers based dreambooth training by JoePenna using prior-preservation loss for 1.000 steps.