Kimi-K3 lora RL day-0 support - #1825
Open
yueming-yuan wants to merge 1 commit into
Open
Conversation
yueming-yuan
requested review from
Shi-Dong,
Zhichenzzz,
fzyzcjy,
guapisolo,
jybsuper,
maocheng23 and
yushengsu-thu
as code owners
July 27, 2026 15:55
Contributor
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
Megatron training backend for Kimi K3 plus the colocated SGLang rollout path it needs for RL. Model: KDA (delta-rule) and MLA attention per layer, the attention-residual snapshot bank, and situ activation, under TP/EP/PP/CP. Native LoRA adapters are applied in-model, with shared-A/per-expert-B factors for the 896-expert MoE, and exported to the rollout engines as HF-named chunks over CUDA IPC. Also included: mbridge and megatron_bridge plugins, megatron->HF conversion, MXFP4 pack/unpack plus an MXFP4->BF16 checkpoint tool, and the launchers.
5 tasks
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Blog
check details and experiment results in:
https://www.lmsys.org/blog/2026-07-27-kimi-k3-day0-support
Usage
sglang branch: https://github.com/sgl-project/sglang/tree/sglang-miles-k3
Docker image will be uploaded within 24h.
About
Megatron training backend for Kimi K3 plus the colocated SGLang rollout path
it needs for RL. KDA and MLA attention per layer, the attention-residual
snapshot bank, and native LoRA adapters under TP/EP/PP/CP — with shared-A /
per-expert-B factors for the 896-expert MoE, exported to the rollout engines
as HF-named chunks over CUDA IPC.