Experimental Stable Diffusion 1.5 finetune on this Digimon BLIP Caption Dataset

Results and output are very rough, trained on slightly less than 900 images