DeepSpeed
Add managed_gradient_accumulation for ZeRO stage 0/1
#8184
Merged

Add managed_gradient_accumulation for ZeRO stage 0/1 #8184

tohtana merged 12 commits into master from sfc-gh-truwase/gas_mgmt_zero01
sfc-gh-truwase
sfc-gh-truwase Add managed_gradient_accumulation for ZeRO stage 0/1
00800f7a
sfc-gh-truwase sfc-gh-truwase requested a review from stas00 stas00 9 days ago
sfc-gh-truwase sfc-gh-truwase requested a review from tjruwase tjruwase 9 days ago
sfc-gh-truwase sfc-gh-truwase requested a review from tohtana tohtana 9 days ago
sfc-gh-truwase sfc-gh-truwase requested a review from loadams loadams 9 days ago
chatgpt-codex-connector
chatgpt-codex-connector commented on 2026-07-27
stas00
stas00 commented on 2026-07-27
sfc-gh-truwase Guard unmanaged gradient accumulation against unsupported features
e7592ea5
sfc-gh-truwase
sfc-gh-truwase Doc gas mgmt
dbe79089
sfc-gh-truwase
sfc-gh-truwase Merge branch 'master' into sfc-gh-truwase/gas_mgmt_zero01
4c6ae671
sfc-gh-truwase
chatgpt-codex-connector
chatgpt-codex-connector commented on 2026-07-28
sfc-gh-truwase Docs: mark unmanaged ZeRO-2/3 and offload as not yet available
c6ba3415
stas00
sfc-gh-truwase Docs: make unmanaged gradient accumulation example use scale_wrt_gas=…
dec2b145
stas00
stas00 commented on 2026-07-29
tohtana
tohtana requested changes on 2026-07-29
sfc-gh-truwase Docs: make loss averaging explicit in unmanaged accumulation example
ab6e8ac7
sfc-gh-truwase Address tohtana review: reject set_gradient_accumulation_boundary/off…
bee8ed56
sfc-gh-truwase Tersify comments in unmanaged gradient accumulation tests
0191153d
sfc-gh-truwase Reject ZeRO overlap_comm in unmanaged gradient accumulation
6b3d3564
sfc-gh-truwase
sfc-gh-truwase sfc-gh-truwase requested a review from tohtana tohtana 5 days ago
tohtana
tohtana requested changes on 2026-07-31
stas00
sfc-gh-truwase
sfc-gh-truwase Track global_samples by actual micro-batch count in unmanaged mode
43070642
sfc-gh-truwase Merge branch 'master' into sfc-gh-truwase/gas_mgmt_zero01
77fbb461
tohtana
tohtana approved these changes on 2026-08-02
tohtana tohtana merged 7a31fe66 into master 3 days ago
tohtana tohtana deleted the sfc-gh-truwase/gas_mgmt_zero01 branch 3 days ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone