DeepSpeed
fix(zero): Ensure full gradient reduction for Muon optimizer with reduce_scatter
#7878

Open

fix(zero): Ensure full gradient reduction for Muon optimizer with reduce_scatter #7878

nathon-lee wants to merge 6 commits into deepspeedai:master from nathon-lee:fix_cp_7807

fix: Ensure full gradient reduction for Muon with reduce_scatter

1dc41225

Update stage_1_and_2.py

a873854e

Fix ZeRO stage to choose BF16 optimizer in test (#7803)

f6ddd754

Update stage_1_and_2.py

15996a95

Merge branch 'deepspeedai:master' into fix_cp_7807

196c7ae5

Merge branch 'deepspeedai:master' into fix_cp_7807

4665aa93

nathon-lee requested a review from

tjruwase 11 days ago

nathon-lee requested a review from

tohtana 11 days ago

chatgpt-codex-connector commented on 2026-02-27

tjruwase requested a review from

PKUWZP 8 days ago

nathon-lee changed the title ~~Fix cp 7807~~ fix(zero): Ensure full gradient reduction for Muon optimizer with reduce_scatter 4 days ago

Reviewers

tjruwase

tohtana

PKUWZP

Assignees

No one assigned

Labels

None yet

Milestone

No milestone