Weavile LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. Claw color can be inconsistent, they can come out black, white or red. Getting a specific claw color may require prompting. Her forehead gem thingy can be duplicated sometimes, which may require inpainting to fix. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on an all female dataset.
