implementation of One-step Diffusion with Distribution Matching Distillation
Model/Pipeline/Scheduler description
https://github.com/Zeqiang-Lai/OpenDMD
Open source status
- [X] The model implementation is available.
- [X] The model weights are available (Only relevant if addition is not a scheduler).
Provide useful links for the implementation
No response
Cc: @kashif maybe you would be interested to check it out.
for sure! let me have a look!
@kashif could you please involve me in as well, would be really happy to help you with this. Also, I am new to community, hence would get to learn as well. Thanks. BTW, I have already read the paper, as well as excavated the corresponding code repo OpenDMD
<3
@kashif Hello, I have very little experience with AI (or at lest insides of it), but I would love to join if I would be of any help.
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.
beep boop
Hi all, we have been independently working on the implementation of this paper as well with @abyildirim. Our primary focus was EDM models and not text-to-image models, and it seems that @Zeqiang-Lai have covered that part 😄. I'm also writing here to note the unet/EDM implementations of the one-step distillation (colab notebook available). I provide the detail below, particularly for the maintainers want to implement this in the library. (Note: I don't know the development status).
Model/Pipeline/Scheduler description
https://github.com/devrimcavusoglu/dmd
Open Source Status
[X] The model implementation is available. [X} The model weights are available (Only relevant if addition is not a scheduler).
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.