lipeirannju / dreambooth-with-colossalai

DreamBooth by colossalai

DreamBooth is a method to personalize text2image models like stable diffusion given just a few(3~5) images of a subject.

• The model used in the experiment

        The base model is CompVis/stable-diffusion-v1-4.

• The dataset employed

        The images are from a British short hair cat of my friends, Ye Tong.

• Parallel settings (if any)

        This experiment is conducted on a 4-V100 machine of a GPU cloud server.

        The parallel setting is the same as defaults. World_size = 4.

• Instructions on how to run your code

        When you want to train the model:

bash ./colossalai.sh > train.log 2>&1

        If you have trained the model, and want to do inference:

python inference.py > inference.log 2>&1

• Experiment results

        The log of the experiment is named as train.log/inference.log

        The generated image is named as output.png

        The image below is an example for inference (Its prompt is "a photo of British Short hair Cat.")