hello

How much images have you put on the dataset to train this model , cause images look broke and also i think tagging or prompt is not proper here i guess , use the tagging website to auto tag the images to get better results , use kohya sd script as trainer