Stable Diffusion 3 support?

Open gpuf4n87 opened this issue 1 year ago • 2 comments

Hi, will it be possible to support Stable Diffusion 3, at least judging by its research paper? If so, could you please update the training code and the models if necessary in the future?

Mar 11 '24 12:03 gpuf4n87

As SD3 has already leveraged T5 as the text encoder, ELLA mainly focuses on CLIP-based models.

Mar 13 '24 12:03 fangyixiao18

If it turns out ELLA is still superior, is it possible you would consider developing something like that for the T5?

Mar 13 '24 18:03 ghost