kiki_moss_v1
LoKr LoRA trained on MOSS Image (int8 SFT base) for the "Studio Ghibli style" look
(trigger word: Studio Ghibli style).
- Network: LoKr, linear 32 / alpha 32, factor 8, full rank
- Final checkpoint:
kiki_moss_v1.safetensors(~4750 steps) - Step saves every 250 steps from 3000 to 4750 included
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support