DeepSpeed
Keep Yuan's o_proj bias whole on every rank and describe it as replicated
#8857

Merged

Keep Yuan's o_proj bias whole on every rank and describe it as replicated #8857

delock merged 3 commits into deepspeedai:master from Achyuthan-S:step6-yuan-oproj-bias

Keep Yuan's o_proj bias whole on every rank and describe it as replic…

bed66b0c

Achyuthan-S requested a review from

hwchen2017 16 days ago

Achyuthan-S requested a review from

loadams 16 days ago

Achyuthan-S requested a review from

tjruwase 16 days ago

Achyuthan-S requested a review from

tohtana 16 days ago

Achyuthan-S requested a review from

copilot-pull-request-reviewer 16 days ago

copilot-pull-request-reviewer commented on 2026-10-18

chatgpt-codex-connector commented on 2026-10-19

Run Yuan's o_proj bias through a real tensor-parallel group

8d1d7fec

delock requested a review from

delock 15 days ago

delock approved these changes on 2026-10-20

Merge branch 'master' into step6-yuan-oproj-bias

566e7176

delock merged b39bf260 into master 13 days ago

Reviewers

delock

hwchen2017

loadams

tjruwase

tohtana

Assignees

No one assigned

Labels

None yet

Milestone

No milestone