Support Skip Layer Guidance (SLG) for Stable Diffusion 3.5 Medium #2590
Conversation
This file contains hidden or 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
LaurentMazare
approved these changes
Nov 1, 2024
|
Thanks, neat addition! |
6 tasks
EricLBuehler
pushed a commit
to EricLBuehler/candle
that referenced
this pull request
Nov 26, 2024
…ggingface#2590) * support skip layer guidance (slg) for stable diffusion 3.5 medium * Tweak the comments formatting. * Proper error message. * Cosmetic tweaks. --------- Co-authored-by: Laurent <[email protected]>
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Outdated suggestions cannot be applied.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.
This closes #2588 .
I noticed SLG doesn't perform well when number of inference steps is low. So, below is a comparison with
--num-inference-steps=50.Appearantly the fingers look better. This could be an evidance that supports claimed benefits (improved anatomy). But other aspects of the image changes, too.
The SLG parameter is copied from that in SD3 reference design, which is originally with dpmpp_2m sampler (not Euler sampler in the Candle example here). So, it might be worth to tweak the parameters with more experiments.