[SD3] Fix mis-matched shape when num_images_per_prompt > 1 using without T5 (text_encoder_3=None) #8558
+10
−2
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Outdated suggestions cannot be applied.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?
This PR is trying to fix the bug when you specified any number greater than 1 in
num_images_per_prompt
when you callStableDiffusion3Pipeline
. An expection occurs when you create the pipeline without T5 text encoder (settext_encoder_3=None
)Reproduction (follow the documentation here):
Bug output
Mitigation:
Bug due to the shape mis-match. For example, in
num_images_per_prompt=4
settting, line 413:will have different shape in
torch.Size([4, 77, 4096])
andtorch.Size([1, 77, 4096])
fix in the function
_get_t5_prompt_embeds
return whentext_encoder_3=None
Before submitting
documentation guidelines, and
here are tips on formatting docstrings.
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.