Repository navigation
fix(vllm_omni): default collective group communication backend - #505
Merged
Merged
Conversation
|
|
lrq619
force-pushed
the
lrq/fix_comm_backend
branch
from
October 12, 2026 07:12
bf0a5ac to
6955ddc
Compare
VllmOmniStrategy.setup_collective_group forwarded *args/**kwargs to the model unchanged, so callers that omitted `backend` got no backend and the group could be created with the wrong communication backend on non-CUDA platforms. Make the signature explicit and fall back to current_platform.communication_backend when `backend` is None, matching the behavior of the other inference strategies. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
lrq619
force-pushed
the
lrq/fix_comm_backend
branch
from
October 12, 2026 13:28
6955ddc to
77a5b9f
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
修复 actor_train（FSDP2）与 actor_infer（vllm_omni）的 device_mapping 不完全相同时，权重同步通信组初始化卡住的问题。该场景需要通过 broadcast 同步权重，但原先 VllmOmniStrategy.setup_collective_group 未补齐 backend 参数，导致推理 worker 无法加入通信组，训练端一直阻塞在 rendezvous。
本次将该方法改为显式签名，增加 backend=None 默认值；未指定 backend 时使用 current_platform.communication_backend，并以关键字参数向模型转发全部六个参数。保留原有 hasattr 判断，worker extension 和调用方无需修改。