transformers
PATCH: add back n-dim device-mesh + fix tp trainer saving
#39693
Merged

PATCH: add back n-dim device-mesh + fix tp trainer saving #39693

S1ro1 merged 18 commits into main from fsdp2-tp
S1ro1
S1ro1 Feat: something
4dd497fc
S1ro1 Feat: initial changes
08f54bbe
S1ro1 tmp changes to unblock
f84ecc45
S1ro1 Refactor
17d2d695
S1ro1 remove todo
56d2c9e2
S1ro1 Merge branch 'main' into fsdp2-tp
622e9b97
S1ro1 Feat: docstring
b35ac20a
SunMarc Merge branch 'main' into fsdp2-tp
33e28196
S1ro1 Merge branch 'main' into fsdp2-tp
83dedd8a
S1ro1 S1ro1 added for patch
HuggingFaceDocBuilderDev
S1ro1 S1ro1 changed the title PATCH: add back n-dim device-mesh PATCH: add back n-dim device-mesh + fix tp hook registration 334 days ago
S1ro1 S1ro1 force pushed from bf21f0a3 to 40fabad8 334 days ago
github-actions
S1ro1 S1ro1 force pushed from 40fabad8 to 83dedd8a 333 days ago
S1ro1 Fix: saving of distributed model in trainer
2423039f
S1ro1 Fix: distributed saving with trainer
4ed16393
S1ro1 Feat: add pure tp saving
b5708c8e
S1ro1 S1ro1 changed the title PATCH: add back n-dim device-mesh + fix tp hook registration PATCH: add back n-dim device-mesh + fix tp trainer saving 333 days ago
ArthurZucker
ArthurZucker commented on 2025-07-28
S1ro1 Only require tp dim if ndim > 1
edd76843
S1ro1 Fix: default to None
d6581d83
ArthurZucker
ArthurZucker approved these changes on 2025-07-28
S1ro1 Fix: better comments/errors
bba981c3
S1ro1 Fix: properly check tp_size attribute
60a96877
S1ro1 Fix: properly check for None in tp_size
354e68f6
S1ro1 Merge branch 'main' into fsdp2-tp
cacd06b3
S1ro1
S1ro1 S1ro1 enabled auto-merge (squash) 333 days ago
S1ro1 S1ro1 merged 4c7da9fe into main 333 days ago
S1ro1 S1ro1 deleted the fsdp2-tp branch 333 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
Assignees
No one assigned
Labels
Milestone