docs: add the Qwen3.8-Flash-Next RL recipe page - #2785
Merged
Merged
Conversation
Qwen3.8-Flash-Next is the GDN + QSA hybrid MoE preview of the Qwen4 architecture, not a variant of the dense Qwen3.8-27B, so it gets its own page beside Qwen3-Next rather than a section under Qwen3.8. The page names the three components that sit outside a stock Megatron layer and the part of the recipe each one forces: hyper-connections in place of the block layernorms, with the hc_head_contraction spec slot they needed in Megatron-LM#89; QSA sparse attention, reimplemented training-side with a Triton forward and backward; and the frozen host-resident PLE table. Then the image, the checkpoint conversion, the launch shapes and the metrics CI gates. Implementation is #2777, still open, pinned together with its SGLang and Megatron branches in docker.io/radixark/miles:qwen38next.
![claude[bot]](https://avatars.githubusercontent.com/in/1236702?s=60&v=4){kind=small}
There was a problem hiding this comment.
Claude Code Review
This repository is configured for manual code reviews. Comment @claude review for a one-time review, or @claude review always to subscribe this PR to a review on every future push.
Tip: disable this comment in your organization's Code Review settings.
Collaborator
|
@claude review always |
Shi-Dong
approved these changes
Aug 27, 2026
yushengsu-thu
approved these changes
Aug 27, 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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Qwen3.8-Flash-Next is the GDN + QSA hybrid MoE preview of the Qwen4 architecture, not a variant of the dense Qwen3.8-27B, so it gets its own page beside Qwen3-Next rather than a section under Qwen3.8.
The page names the three components that sit outside a stock Megatron layer and the part of the recipe each one forces: hyper-connections in place of the block layernorms, with the hc_head_contraction spec slot they needed in Megatron-LM#89; QSA sparse attention, reimplemented training-side with a Triton forward and backward; and the frozen host-resident PLE table. Then the image, the checkpoint conversion, the launch shapes and the metrics CI gates.
Implementation is #2777, still open, pinned together with its SGLang and Megatron branches in docker.io/radixark/miles:qwen38next.