Dedenne LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. Trained on a low image count, inconsistent dataset, expect issues. Tail placement and color can be unstable. Ear color can be unstable. Tends to add hair without prompting for it. Loses a lot of detail when adding clothes to the prompt, you'll likely have to increase the LoRA strength when prompting for heavier clothing. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on a mixed male/female dataset.