Skip to content

fast-stable-diffusion, +25-50% speed increase + memory efficient.

Notifications You must be signed in to change notification settings

digitalblossom/fast-stable-diffusion

 
 

Repository files navigation

fast-stable-diffusion colabs, +25-50% speed increase + memory efficient + DreamBooth

2 Colab adaptations for both hlky AUTOMATIC1111 Webui versions of stable diffusion implementing the optimization suggested by https://github.com/MatthieuTPHR : huggingface/diffusers#532, using the MemoryEfficientAttention implementation from xformers (cc. @fmassa, @danthe3rd, @blefaudeux) to both speedup the cross-attention speed and decrease its GPU memory requirements.

All you have to do is enter your huggingface token only once and you're all set, the colabs will install the repos and the models inside Gdrive, so the loading will be fast everytime you use it, enjoy !!

hlky WEBUI Open In Colab

AUTOMATIC1111 WEBUI Open In Colab

Minimalistic relaxed mode if you just want to have fun Open In Colab

If you encounter any issue or you want to update to latest webui version, remove the folder "sd" or "stable-diffusion-webui" from your GDrive (and GDrive trash) and rerun the colab.

fast-dreambooth colab, +65% speed increase + less than 14GB VRAM, CKPT output.

Train your model using this easy simple and fast colab, all you have to do is enter you huggingface token once, and it will cache all the files in GDrive, including the trained model and you will be able to use it directly from the colab, make sure you use high quality reference pictures for the training. Thanks to https://github.com/jachiam for his diffusers to CKPT script, the trained models now can now be used in SD WEBUI from AUTOMATIC1111 and hlky and others.

DreamBooth Open In Colab

About

fast-stable-diffusion, +25-50% speed increase + memory efficient.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 53.8%
  • Python 42.9%
  • JavaScript 2.0%
  • CSS 1.3%