Hello,
I have a question: training example from examples/dreambooth/train_dreambooth.py
uses DDPM noise scheduler in the training process while default stable diffusion pipeline uses PNDM. What is the reason for that?
Hello,
I have a question: training example from examples/dreambooth/train_dreambooth.py
uses DDPM noise scheduler in the training process while default stable diffusion pipeline uses PNDM. What is the reason for that?
Bumping this. I have the same question.