Trained from Noco's work, based on PonyXL v6
V02, re-did dataset, trained with smaller dim
noco_(adamas)_PonyXL_style_v01.safetensors