DeepSpeed
[AutoEP] Stage expert weight gradients across microbatches
#9114
Merged

[AutoEP] Stage expert weight gradients across microbatches #9114

PKUWZP merged 3 commits into master from autoep-deferred-dw-staging
PKUWZP
PKUWZP PKUWZP requested a review from delock delock 4 days ago
PKUWZP PKUWZP marked this pull request as ready for review 4 days ago
PKUWZP PKUWZP requested a review from tjruwase tjruwase 4 days ago
PKUWZP PKUWZP requested a review from tohtana tohtana 4 days ago
PKUWZP PKUWZP requested a review from hwchen2017 hwchen2017 4 days ago
PKUWZP PKUWZP requested a review from loadams loadams 4 days ago
chatgpt-codex-connector
chatgpt-codex-connector
chatgpt-codex-connector commented on 2026-11-18
PKUWZP PKUWZP force pushed from 6e688506 to f0166f96 4 days ago
PKUWZP PKUWZP added enhancement
PKUWZP PKUWZP force pushed from 9aca13d6 to 4954ced0 4 days ago
PKUWZP PKUWZP requested a review from sfc-gh-truwase sfc-gh-truwase 4 days ago
PKUWZP [AutoEP] Stage expert weight gradients across microbatches
ed819ec1
PKUWZP [AutoEP] Give the route test stubs the staging attributes
799bd499
PKUWZP [AutoEP] Return staged expert gradients in the same backward on every…
6ee0cbb9
PKUWZP PKUWZP force pushed from ea50751b to 6ee0cbb9 3 days ago
PKUWZP PKUWZP added roadmap
delock
delock approved these changes on 2026-11-20
PKUWZP PKUWZP merged d70adc32 into master 1 day ago
PKUWZP PKUWZP deleted the autoep-deferred-dw-staging branch 1 day ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone