Flux2klein models part1#439
Open
amepas wants to merge 1 commit into
Open
Conversation
amepas
force-pushed
the
flux2klein-models-part1
branch
5 times, most recently
from
755a412 to
71f6548
Compare
amepas
force-pushed
the
flux2klein-models-part1
branch
from
71f6548 to
e88aff4
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Part 1 of 2 for Flux.2-klein onboarding.
Contains core Flax model architectures, weight converters, and self-contained modular unit tests:
- Onboarding Qwen3 into JAX+TPU (qwen3_flax.py)
- Flux2-klein Transformer modifications (SwiGLU, AdaLayerNormZero, global modulation)
- RoPE & weight conversion utilities (util.py)
- Modular component unit test suite (generate_flux2klein_test.py)
This code does not affect any other models implemented in max diffusion.