the effect it has on outputs is amazing, do you mind telling me the lora training settings used on this one? (steps, unet, dim, ect)