search

yandex-research / ddpm-segmentation

Label-Efficient Semantic Segmentation with Diffusion Models (ICLR'2022)
https://yandex-research.github.io/ddpm-segmentation/
MIT License
662 stars 60 forks source link

How can I train a DDPM model on my own dataset? #15

Closed Liareee closed 1 year ago

Liareee commented 1 year ago

Hi! Thank you very much for sharing the code, but I'm a newbie in Diffusion Model. I know that the FFHQ-256 is trained by yourself. I would like to try something on other dataset, but I don't know how to do. Can you please share code about how to train a pretrained model on my own dataset ( like CIFAR10 )?

Thank you very much!

dbaranchuk commented 1 year ago

There are many pretrained models around and corresponding high-quality implementations. Please refer to https://github.com/openai/guided-diffusion, https://github.com/openai/improved-diffusion, https://github.com/ermongroup/SDEdit.

For CIFAR10, you can consider https://github.com/NVlabs/edm