SDXS-1024 Release?

[ApolloRay]

No description provided.

[IDKiro]

I like open source, but because I am no longer a Ph.D. student, based on commercial considerations, this may not be a simple matter. I will do my best to provide more, including more pretrained weights and training code.

[ApolloRay]

I like open source, but because I am no longer a Ph.D. student, based on commercial considerations, this may not be a simple matter. I will do my best to provide more, including more pretrained weights and training code.

It would be best to provide a version of the training code for the sdxl model. Too many repos now are based on sd1.5.

[ApolloRay]

I like open source, but because I am no longer a Ph.D. student, based on commercial considerations, this may not be a simple matter. I will do my best to provide more, including more pretrained weights and training code.

I am very concerned about the latest model progress. If the code can be provided, maybe I will place an order for Xiaomi su7. HHHHH

[AMEERAZAM08]

https://huggingface.co/spaces/ameerazam08/SDXS-GPU-Demo

@IDKiro let me know if anything need to change in this I have created ite running on A Series GPU

[wibur0620]

I hope comfyui will support sdxs soon

[IDKiro]

https://huggingface.co/spaces/ameerazam08/SDXS-GPU-Demo

@IDKiro let me know if anything need to change in this I have created ite running on A Series GPU

Thank you very much for your contribution, SDXS is still in pre-release, so we don't provide demos for the time being, we will mention your demo in the README of hugging face model.

[311-code]

Yes I am also mostly interested in the 1024 version also.

Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

[ApolloRay]

Yes I am also mostly interested in the 1024 version also.

Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

xiaomi company

[IDKiro]

Yes I am also mostly interested in the 1024 version also.

Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

This is just a research project and the related technology may not land on Xiaomi's products. However, it's not up to me to make the relevant technology public to peer companies, so I'd say it's not a simple matter.

[ApolloRay]

Yes I am also mostly interested in the 1024 version also.

Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

You can try OpenDMD https://github.com/Zeqiang-Lai/OpenDMD，I will try it in this week.

[ApolloRay]

Yes I am also mostly interested in the 1024 version also.
Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

This is just a research project and the related technology may not land on Xiaomi's products. However, it's not up to me to make the relevant technology public to peer companies, so I'd say it's not a simple matter.

If you can release the training code, it will be better.

[IDKiro]

Yes I am also mostly interested in the 1024 version also.
Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

You can try OpenDMD https://github.com/Zeqiang-Lai/OpenDMD，I will try it in this week.

Nice repo, I'll refer to it as well.

[IDKiro]

Yes I am also mostly interested in the 1024 version also.
Can you provide more detail on the 1024 holdup? I know you mentioned commercial reasons. Is this because of stability ai license? I always thought training and releasing was fine for projects like these, but maybe there's something I don't know.

You can try OpenDMD https://github.com/Zeqiang-Lai/OpenDMD，I will try it in this week.

I just read this paper's figures, which has a one-step model training method that is closer to the idea of this paper, and it looks like I need to add it as a citation in the revised version of the paper.

[IDKiro]

The subsequent open source program has been terminated by the Company. I'll reopen these issues if there are subsequent progress.