@juliandmc2001844 Good to hear. I posted some images to your models to help fund your training.

@juliandmc2001844 Not an issue and I didn't even notice it. My advice on eye color: Unless the color is so unique/specific its not a booru tag, its best to just not bother tagging the eye color in training and just tag it in generation. The eye shape will get learned as part of the character's trigger and 1girl (same way it learns chin, nose etc.). Eye color can often vary to a ridiculous degree even in a single series so its often a waste, and it has such a minor impact (low percent of image) on training's attempts to match the training data (which already doesn't see color as we do on its primary method of vision) the tokens.

Also while I'm mentioning this: bangs isn't a tag, despite what the some auto-label solutions try to claim. For characters you really only need one hairstyle tag most of the time unless its something you know Illustrious (etc.) won't understand at all without explicitly being told (mainly certain heavy asymmetry, beyond just bangs, that's not already implied by a hairstyle tag like "side ponytail"). One tag per article of clothing (+pattern+asymmetry) is generally enough for that too.