https://huggingface.co/spaces/ysharma/Low-rank-Adaptation

Duplicate the space and go to settings to assign a gpu to train your own model

huggingface is like github so you can clone the space and run it locally as you would a github repo

git clone https://huggingface.co/spaces/ysharma/Low-rank-Adaptation

pip install -r requirements.txt

python app.py

Did you do an evaluation of the quality impact?

Yep, I'm curious to see too

iirc its not as good as normal dreambooth because it cant train the text encoder.

Constantly gives me an error message. When trying to run a duplicate, it says I need to buy GPU time.

now it can train both, unet and textencoder

it seems it can now