Is there a plan to update to v0.15.1 for diffusers?

Run into this bug which was fixed in that release

I tried to update but looks like it is done though a manual git clone step so assuming there must be reasons to not using pip

Thank you very much

Hey @shao

Yes, actually, 0.15.1 and a few other patches have been committed to the ‘dev’ branch, BUT, unfortunately automated tests didn’t pass so no build is available yet.

You can however build this yourself. All the integration tests pass for me locally, I need to investigate why 1 test is failing on the cloud but it seems to be network related and nothing that should impact real usage.

I should have a chance over the weekend to try sort this out if you want to wait.

As for why we use git checkout over pip:

  • we can pin to particular commits outside of the official release cycle
  • we get point-in-time copies of community contributions which aren’t included in the official releases.

Hope that helps!

yeah, super helpful. I will test out the dev branch. Thanks

Great. Seems I did break one or two things with the very latest commits though, looking at it now and will update :sweat_smile:

All fixed and tests passing :slight_smile:
Latest dev build pushed to docker hub.

So you can now docker run ... gadicc/diffusers-api:dev or use it in your own Dockerfile with FROM gadicc/diffusers-api:dev. (Pretty sure you know this already but am writing for the sake of others :)).

Oh sorry, last thing… obviously it’s still a dev release and not an official release yet. Tests are passing but YMMV. It might be a while until I make the next official release as I have a few other things going on and want it to get a bit more real world usage first. Having said that, once I finish up a few other things I’ll probably start using the dev release on soon…ish :slight_smile:

No worries. Thank you so much for what you have already done.

I have to say again: the code quality is solid solid in this repo!

1 Like

Thanks for the kind words. It’s getting there :sweat_smile:

This is actually my first time working in Python so there’s been a bit of a learning curve. But was very happy to get automated testing and semantic releases in place… we still have relatively low code coverage in unit tests but it’s been great to be able to refactor stuff and know that all the e2e tests are still passing. So, definitely some fun times ahead :slight_smile:

Thanks for your support!

1 Like