SD3 Gui Sampling + lora extraction

[KujoAI]

Hi there bmaltais, I've trained a dream booth model for sd3, but I noticed that it doesn't sample any images in the gui it just skips them.

also I was wondering if there is any code yet to extract the lora from the db model?

Thank you for continuing to work and providing us with the gui to train! have a nice day!

[araleza]

There's no official support for SD3 sampling while training yet, but if you're okay with modifying a local sd-scripts code file, this will make most basic sampling features work:

kohya-ss/sd-scripts#1374 (comment)