Trained using anime screencaps and fanart on the animefull-final model for 13 epochs. I had to use quite a high network dim of 128 so it could understand the snake part better, which I resized to one of 64 using the resize script.
Use Miia in your prompt, I'd also recommend using facial mark to get those triangles on her face, I tried baking them into the character itself but the AI just wouldn't pick up on it unless specifically prompted.
Since it's trained on the animefull-final model it works well with most derivative merges of it such as the Anything and AOM series, however unlike my other character LoRAs this has a lot of trouble with realistic models such as ChilloutMix.
