I pruned the dataset and retrained on animefull. Funnily enough, the style transfer is more faithful, but you may need to modify your prompting to get it to work on your Clip Skip 2 model of choice. Overall, it works pretty well on AnyLoRA.
I pruned the dataset and retrained on animefull. Funnily enough, the style transfer is more faithful, but you may need to modify your prompting to get it to work on your Clip Skip 2 model of choice. Overall, it works pretty well on AnyLoRA.