ClashSAN

i don't understand.

Is there already an option for it ? I can't remember. .jpg has exif comment, maybe you can use that. 

is this issue resolved?

hi @Annuakin, I am able to train both textual inversion and LoRA's (unet only) in 4gb, using https://github.com/kohya-ss/sd-scripts I suggest you use this repo instead, as it has 8bit training...

yes, ~25 min. it depends if you can use bitsandbytes and xformers well on your machine.

--skip-torch-cuda-test --precision full --no-half --use-cpu all

is this issue resolved?

1. possibility add `fastapi==0.90.1` to `requirements.txt` in the webui folder. then double click webui-user.bat you can remove it after running once, this would install the correct fastapi version. 2. possibility...

you might have to rebuild the venv or reinstall i guess. this doesn't show on a fresh install

any screenshots?