v1 - was too overcooked and destructive
v2 - This lora trained on the same dataset but with rank 32, low noise bias and 1.5K steps only.
Resulting in much better preservation of z image inherent capabilities
내가 좋아하는 스타일로 많은 waifu 애니메이션 이미지로 학습됨.
랭크 128
스텝 15000
매우 과하게 조리되었을 가능성이 높음.
