Remove unneeded flux2 text encoder float cast before quantization - #135
Merged
Merged
Conversation
tjia1818
approved these changes
Jul 31, 2026
stikves
approved these changes
Jul 31, 2026
carinapeng
approved these changes
Jul 31, 2026
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
4 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
In the FLUX2 export pipeline, we forced the text encoder to be float32 cast before compression/conversion. This was done to ensure text encoder parity during earlier seed builds, but it is no longer needed. We still get high parity and good image quality without this cast.
Motivation
Removing this allows the text encoder to default to float16 (like the other components) during this part of the export. This drastically reduces the memory pressure during model load/runtime. On a MacBook, the peak memory footprint dropped by almost 50%. On an iPhone/iPad, it also drops a significant amount.
Validation
Text encoder parity looked good, and image quality remained high.
Before this PR:
With this PR:
