Trained on 20 images, 2000 steps.
With TheLastBen fast-stable-diffusion (https://github.com/TheLastBen/fast-stable-diffusion)
use the token tvdino
