v0.1

this repo contains some heavily experimental sd3.5M anime finetunes, again, this is very experimental and will probably not be good yet, here are some checkpoints for testing and / or feedback

current todos:

get nlp captions
multicaptioned dataset
find the hparams

if you do use this model, the tag list is wd-tagger-eva and uses this, prompting format, 1girl, and if you do use this, please give feedback in the discussion tab or in my (raiko.horikawa) discord dms