Skip to content

Dreambooth x Brev: open a pre-configured and ready-to-go DreamBooth GPU environment

License

Notifications You must be signed in to change notification settings

brevdev/dreambooth

 
 

Repository files navigation

Much of the code in this implementation was borrowed from Shivam Shrirao. Huge thanks to him!

We've pre-configured a Brev environment that'll run Dreambooth out of the box! To get started, hit this link to create a Brev environment.

1) Open your Dreambooth environment:

Open your new DreamBooth environment in VSCode:

brev open dreambooth --wait

If you don't have the Brev CLI installed, you can install it here.

2) Upload training data:

  1. Upload about 20 or so photos of someone you want to generate SD samples to the directory (you can drag and drop to vscode)
  2. Inside launch.sh change instance_data_dir to point to your training data (line 12)
  3. Then change the prompts you use on lines 13 and 22 of launch.sh e.g. "a photo of Jeremy" & "a photo of Jeremy wearing sunglasses"

3) Run the training job:

Setup your HuggingFace token:

huggingface-cli login

It'll prompt you to add your huggingface token (make sure you've accepted the Hugging Face license agreement).

Then run the training job :

sh launch.sh

(this should take about 5 minutes)

4) Generate samples:

To then do your own inference run:

conda activate diffusers
python inference.py "fine-tuned-model-output/800" "a photo of sks dog wearing sunglasses"

To check out all the things you can do, take a look at Shivam's example

About

Dreambooth x Brev: open a pre-configured and ready-to-go DreamBooth GPU environment

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.4%
  • Shell 4.6%