Lora question
#31
by
vodll
- opened
So the z-image-turbo is the distilled version and we're waiting for the base model to be able to fine tune.
Will loras created on the base model work on the distilled version? I'm guessing yes because Chroma has something similar where if you train on the base model the loras work just fine on any of the distilled flavors.
AI Toolkit - "Add support for training Z-Image Turbo with a de-distill training adapter"
https://github.com/ostris/ai-toolkit/commit/4e62c38df5eb25dcf6a9ba3011113521f1f20c10