Audie LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. Works great, main issue is an extreme bias towards teal for dresses due to every dress in the dataset being teal. Look at my cocktail dress photo prompt for an idea of how to combat that. Add (sunglasses) to your prompt to add her signature head sunglasses. Has trouble getting sunglasses on her eyes due to them always being on her head in the training data. Add (pineapple dress) to your prompt to get her signature dress. Should have wide compatibility with model mixes based on NovelAI, which is most of them.