v1 Pony - Trained on 76 images from Crisis-Omega.

v2 iLL - 275 images trained from e621.

v2 I'm experimenting with chatgpt to help me quickly get e621 tags + images, without training on unnecessary tags like "what" or "lol_comments"