-
Notifications
You must be signed in to change notification settings - Fork 4.9k
Pull requests: deepspeedai/DeepSpeed
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
fix: stop DeepSpeedConfig writing max_grad_norm back into the caller's config dict
#8289
opened Aug 22, 2026 by
ebarkhordar
Contributor
Loading…
fix(engine): resolve inference workspace attribute lookup
#8288
opened Aug 22, 2026 by
nathon-lee
Contributor
Loading…
docs: note async cpu_checkpointing perf and expandable_segments
#8287
opened Aug 22, 2026 by
sfc-gh-truwase
Collaborator
Loading…
1 of 2 tasks
Fix ZeroDivisionError in compute_elastic_config return_microbatch on non-0.2 elasticity
#8286
opened Aug 21, 2026 by
ebarkhordar
Contributor
Loading…
Register native pinned host memory with CUDA for GPU DMA
#8283
opened Aug 20, 2026 by
sfc-gh-truwase
Collaborator
Loading…
3 tasks done
Enable optimized Adam backend for MuonWithAuxAdam optimizer
#8278
opened Aug 19, 2026 by
jinyouzhi
Contributor
Loading…
Fix ZeRO parameter alignment for grouped_mm
#8277
opened Aug 19, 2026 by
fwerkor
Contributor
Loading…
Reject invalid ZenFlow ratio and update interval boundaries
#8274
opened Aug 18, 2026 by
tandede
Loading…
fix: Only bind device id when needed, Fixes #8248
#8269
opened Aug 17, 2026 by
pengdurice
Contributor
Loading…
Make the WarmupCosineLR ratio flags reach the config
#8268
opened Aug 17, 2026 by
vineethsaivs
Contributor
Loading…
Fix comms logger KeyError when log_name is omitted
#8267
opened Aug 17, 2026 by
jinyouzhi
Contributor
Loading…
Fix Triton NFS detection crash when df wraps long device names
#8259
opened Aug 16, 2026 by
atirna
Loading…
Route ZeRO/SuperOffload pin sites through accelerator pin_memory
#8256
opened Aug 15, 2026 by
sfc-gh-truwase
Collaborator
Loading…
4 tasks done
Fix repeated FlopsProfiler metric accumulation
#8246
opened Aug 12, 2026 by
baremetaldevx86
Contributor
Loading…
Drop the documented grad_hooks ZeRO option, which does not exist
#8242
opened Aug 10, 2026 by
vineethsaivs
Contributor
Loading…
[AutoTP] Replace tp_shard process-wide globals with per-model AutoTPMeta
#8241
opened Aug 10, 2026 by
delock
Collaborator
Loading…
Filter --include against the real slots, not against itself
#8239
opened Aug 9, 2026 by
vineethsaivs
Contributor
Loading…
Add adaptive gather-buffer pooling for DeepCompile ZeRO-3
#8234
opened Aug 7, 2026 by
tohtana
Collaborator
Loading…
Previous Next
ProTip!
Adding no:label will show everything without a label.