Ninetales LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. There are a lot of tails to deal with, and they don't always end up in the right places. I recommend lowering the LoRA strength if you can't wrangle the tails. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on an all female dataset.
