Update dataset. Retrain the model using new craft.
Reduce loss to 0.09. Adjust rank to 32 to balance file size and quality.
Use offset noise to improve generating ability for lighter/darker images. Reduce model file size to ~1MB. The preview images are generated using latest version of YesMix.
A LORA model trained with ~110 images of Signora from Genshin Impact. The 784mb VAEs (NAI, Orangemix, Anything, Counterfeit) are recommended. 0.6~0.8 weights should be enough.