download_from_original_stable_diffusion_ckpt initializes correct default pipeline for SDXL #5784
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Previously the
download_from_original_stable_diffusion_ckptfunction, when run with a SDXL model, would default to aStableDiffusionPipelinepipeline class. This would lead to downstream errors because the SDXL model has atext_encoder_2which is not recognized in theStableDiffusionPipeline.A few lines below, the function checks if the model is a SDXL or SDXL-Refiner and sets the
model_typevariable accordingly.I moved the initialization of the default pipeline (when None is set specifically) below this detection if the model is SDXL or SDXL-Refiner and initialize the correct default pipeline. This makes it easier for users who for example run the
convert_original_stable_diffusion_to_diffusers.pyscript.Before submitting
This issue came to my attention only a few hours ago and I immediately fixed it but it was part of a discussion at Issue#4488. It basically adds a better default to MR#4461
documentation guidelines, and
here are tips on formatting docstrings. No changes needed
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@patrickvonplaten @yiyixuxu