Benutzer-LoRA
Trained from Choco's work, based on PonyXL v6.V02, re-did dataset, trained with smaller dim
