DeepSpeed
Fix shared loss gradient accumulation
#8245
Merged

Fix shared loss gradient accumulation #8245

tohtana
tohtana tohtana requested a review from tjruwase tjruwase 59 days ago
tohtana tohtana requested a review from loadams loadams 59 days ago
chatgpt-codex-connector
chatgpt-codex-connector commented on 2026-08-11
pengdurice
sfc-gh-truwase
sfc-gh-truwase approved these changes on 2026-08-20
Fix shared loss gradient accumulation
9ec5deb9
tohtana tohtana force pushed from a99d90c5 to 9ec5deb9 43 days ago
tohtana
tohtana tohtana merged 4ffb4e5c into master 43 days ago
tohtana tohtana deleted the tohtana/fix-zero2-bf16-shared-grad-8224 branch 43 days ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone