Fixes for --flux_fast_schedule - #803
Merged
Merged
Conversation
…and simplify the code. Also fixing the sigma schedule to what should be correct based on SD3-Turbo paper.
Owner
|
would you be interested in attempting to port the assistant lora code from ostris' ai-toolkit? |
Contributor
Author
|
Not really interested and I don't have enough spare time either. |
Owner
|
same :D |
samzhang996-hue
pushed a commit
to samzhang996-hue/SimpleTuner
that referenced
this pull request
Apr 27, 2026
Fixes for --flux_fast_schedule
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 fixes issue #780. I tested it with a batch size of 2. I'm also fixing the schedule to what I believe is correct based on SD3-Turbo paper. Schnell training is still broken based on a quick 500 step test run. So the advice is still to stay away from Schnell.