Fix bug in stable diffusion when mask_pad_tokens is false #147
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
coryMosaicML
added a commit
to coryMosaicML/diffusion
that referenced
this pull request
Jun 4, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Currently when
mask_pad_tokens=Falsein the stable diffusion model, the mask is still passed to the text encoder at generation time. This introduces a train/test mismatch because the masked tokens in the text encoder output are different from the unmasked tokens, and the model is trained with the unmasked pad tokens. The result is a degradation in image quality. The fix is to only use the mask from the tokenizer output ifmask_pad_tokens=True.