Tinkaton LoRA trained on NovelAI with Kohya training scripts using the DAdaptAdam optimizer at 512x512 resolution. Works pretty well, though you might want to turn down the LoRA strength if you run into issues. I tried to get the hammer added as a trigger word and partially succeeded. Adding (hammer, holding hammer) will attempt to give the little menace her hammer, though it ends up just being the hammer shaft most of the time. It can take some tries, but it's possible to get a picture with the hammer in it. Adding (furry, nose, tail) to my negative improved consistency on YiffyMix, as well as adding (noseless, long hair) to my positive. Should have wide compatibility with model mixes based on NovelAI, which is most of them. Trained on an all female dataset.

Files with SHA256: a7ae379c66702f24d92cb8deb334936e3536ecc789cf5aed2cf1b8706c630425