(Nana Asteria 0.5 strength)
>>59576638Thank you. I'll see what I can do tomorrow. I think I still have the datasets ready so it shouldn't be that hard to retrain with better settings.
I am interested in the archival part of things as some of them like Aruru which I linked a few days ago don't have their content up anymore so they're pretty much 'lost media' if anyone else wants to train them nowadays unless we get a better way to train models with just 1-4 images. And with Twitter starting to delete inactive accounts even more might be lost in the future. I would love to help preserve as much as possible.
You're doing AIGod's work, anon!
