DeepSpeed
TiledMLP + SequenceTiledCompute: improve the bs>1 use-case
#7422
Merged
Overview
Commits
17
Changes
View On GitHub

TiledMLP + SequenceTiledCompute: improve the bs>1 use-case #7422

stas00 merged 17 commits into master from stas/tiled-mlp-bs-gt-1
stas00
stas00 [TiledMLP]: fix for bs>1
1ce72101
stas00 copy the fix to other tilers, improve tests
fba09c19
stas00 more solid test
4babd618
stas00 [TiledMLP] improve work with bs>1
0fa79f1d
stas00 Merge remote-tracking branch 'origin/master' into stas/tiled-mlp-bs-gt-1
f85c7a02
stas00 stas00 changed the title Stas/tiled mlp bs gt 1 TiledMLP: improve the bs>1 use-case 1 year ago
stas00 copy solution
1242e5d8
stas00 cleanup
42d92447
stas00 cleanup
527ed311
stas00 stas00 changed the title TiledMLP: improve the bs>1 use-case TiledMLP + SequenceTiledCompute: improve the bs>1 use-case 1 year ago
stas00 optimize
faea42b5
stas00 sync
47f0b68f
stas00 flatten bs+seqlen
d6e57f13
stas00 Merge remote-tracking branch 'origin/master' into stas/tiled-mlp-bs-gt-1
612a2b03
stas00 undo for TiledCompute - need test for logits/loss
2e2fd31e
stas00 Merge branch 'master' into stas/tiled-mlp-bs-gt-1
d0045ed8
stas00 stas00 marked this pull request as ready for review 1 year ago
stas00 stas00 requested a review from tjruwase tjruwase 1 year ago
stas00 stas00 requested a review from tohtana tohtana 1 year ago
stas00 stas00 requested a review from loadams loadams 1 year ago
stas00 Merge branch 'master' into stas/tiled-mlp-bs-gt-1
b3c034e5
loadams Merge branch 'master' into stas/tiled-mlp-bs-gt-1
6dc52fc4
loadams Merge branch 'master' into stas/tiled-mlp-bs-gt-1
34b983d8
loadams
loadams approved these changes on 2025-07-16
stas00 stas00 merged c2bb53f2 into master 1 year ago
stas00 stas00 deleted the stas/tiled-mlp-bs-gt-1 branch 1 year ago
stas00

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Terms Of Use Privacy Policy Refund Policy Impressum