Open pedr0sorio opened 1 month ago
Hi Pedro,
Thank you for your interest in MaskDiT! Adapting the MaskDiT code for the diffuser library should be straightforward for the inference part, as it's just the standard EDM sampler. The training part might take a bit more time to adapt. Unfortunately, we don't currently have the bandwidth to do it immediately. Nevertheless, for research purposes, the code in this repository should be easy to use and adapt for your own projects.
Additionally, you can also add a request to reopen the diffuser github issue of incorporating MaskDiT. HuggingFace members might be able to help with that as well.
Dear MaskDiT Team,
I recently read your paper and found your work impressive. Congratulations on this remarkable achievement! I am interested in trying out your solution as well as incorporate it in a few of my projects. It would be beneficial to the community if the model was available on Hugging Face. In addition, it would be convenient to have the code base incorporated in the diffusers framework. Do you plan on doing any of that?
Nevertheless, great stuff! Best, Pedro