-
Notifications
You must be signed in to change notification settings - Fork 5k
Pull requests: deepspeedai/DeepSpeed
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Run Muon once per optimizer step under ZeRO-3, not once per micro-batch
#8600
opened Sep 19, 2026 by
alanhuangyoo
Contributor
Loading…
docs: remove duplicated words and garbled fragments in comments and docs
#8599
opened Sep 19, 2026 by
simpleqt
Loading…
docs(inference): align fp_quantize's docstring with its signature
#8598
opened Sep 19, 2026 by
simpleqt
Loading…
docs(autotp): point the supported-models link at the readthedocs page
#8597
opened Sep 19, 2026 by
simpleqt
Loading…
docs(comms-logging): fix the 'deepspeed.com.log_summary' module name
#8596
opened Sep 19, 2026 by
simpleqt
Loading…
docs(megatron): fix the stale get_cuda_rng_tracker module path
#8595
opened Sep 19, 2026 by
simpleqt
Loading…
docs: fix case-mismatched ZeRO tutorial links in the ZeRO-3 Offload post
#8594
opened Sep 19, 2026 by
simpleqt
Loading…
Check input + max_new_tokens against max_out_tokens in InferenceEngine.generate
#8592
opened Sep 19, 2026 by
udsy19
Loading…
Decay OneCycle momentum instead of growing it past 1.0
#8589
opened Sep 18, 2026 by
vineethsaivs
Contributor
Loading…
Report the unscaled gradient norm from FP16_UnfusedOptimizer
#8588
opened Sep 18, 2026 by
vineethsaivs
Contributor
Loading…
Weight the TiledLoss backward pass the way the forward does
#8587
opened Sep 18, 2026 by
vineethsaivs
Contributor
Loading…
Fix: Replace assert statements with ValueError in runtime config validation
#8585
opened Sep 18, 2026 by
mayuriphad
Loading…
fix(rollout): reject shared prefill with continuous batching
#8583
opened Sep 18, 2026 by
modelpath-dev
Loading…
[AutoTP] Implement VocabParallelEmbedding and enhance tied model support
#8582
opened Sep 18, 2026 by
jinyouzhi
Contributor
Loading…
Release AutoEP router-logit caches on checkpoint replay exits
#8581
opened Sep 18, 2026 by
yh0903
Contributor
Loading…
6 tasks done
[muon] Per-head Newton-Schulz for fused QKV projections
#8580
opened Sep 18, 2026 by
0z5a
Contributor
Loading…
Accept torch.int8 in init_inference's dtype validation gate
#8578
opened Sep 18, 2026 by
udsy19
Loading…
Describe the GPTBigCode and Yuan shared-QK layouts
#8575
opened Sep 17, 2026 by
Achyuthan-S
Contributor
Loading…
Keep runtime configuration validation active under python -O
#8573
opened Sep 17, 2026 by
Saisharathchandranandnetha
Loading…
Fix memory leak and NULL-deref in shm_initialize
#8565
opened Sep 17, 2026 by
tritsystem
Loading…
5 of 6 tasks
AutoEP: share one DeepEP buffer across MoE layers instead of one per layer
#8557
opened Sep 16, 2026 by
yh0903
Contributor
Loading…
Previous Next
ProTip!
Type g p on any issue or pull request to go back to the pull request listing page.