Add option to disable normalize embeddings after after calculating emphasis. #14874
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
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Outdated suggestions cannot be applied.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.
Description
In SDXL, there have been reports of artifacts occurring when combining prompts or using long prompts. I have discovered the cause of this phenomenon. It is the normalization that occurs after the calculation of emphasis. During this normalization, artifacts arise due to division by values close to zero and the inversion of signs. This normalization process dates back to the SD1.X era, where it seems to work well. However, in certain models of XL, it appears to have the opposite effect. Therefore, I will add an option to toggle this normalization on and off.
The detailed explanation can be found here or here(Japanese)
Checklist: