Skip to content
New pull request New
51 Open 965 Closed
51 Open 965 Closed
Author
Filter by author
Label
Filter by label
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Milestones
Filter by milestone
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀
feat: Support for training with MoE module in PipelineEngine
#1942 opened May 7, 2022 by shjwudp Loading…
[bug] Add user-defined launcher args for MPI launcher
#1933 opened May 2, 2022 by mrwyattii • Draft
fix: remove layer_past storing in DeepSpeedTransformerInference
#1930 opened May 1, 2022 by codertimo Loading…
1
Fairseq support
#1915 opened Apr 26, 2022 by jeffra Loading…
7 tasks done
Fix: Sparse tensors not updating
#1914 opened Apr 26, 2022 by Dipet Loading…
6
Fix DeepspeedDataLoader length
#1908 opened Apr 22, 2022 by lkm2835 Loading…
Configure fused fp16 mode
#1882 opened Apr 6, 2022 by tjruwase Loading…
Bug Fixes for autotuner and flops profiler
#1880 opened Apr 5, 2022 by cli99 Loading…
Add checkpoint_comm_enabled flag
#1778 opened Feb 16, 2022 by jeffra Loading…
1
Modify unit test to cover more cases
#1775 opened Feb 15, 2022 by RezaYazdaniAminabadi Loading…
DataLoader Length Fix
#1718 opened Jan 23, 2022 by Sanger2000 • Draft
Add RTS and token masking to top-2 gating + configurable jitter epsilon
#1700 opened Jan 17, 2022 by ykim362 • Draft
1
remove force-multi and fix None val check in base tuner in autotuning
#1657 opened Dec 22, 2021 by cli99 Loading…
5
[zero-3] add support for new params added during fwd pass
#1606 opened Dec 1, 2021 by jeffra Loading…
8
support batch size dimension in 2D sparse attention mask
#1597 opened Nov 29, 2021 by jglaser Loading…
1
Tweaks for lm-eval-harness
#1592 opened Nov 26, 2021 by zphang Loading…
1
Optimizer state loading fix for bitsandbytes 8-bit optimizers.
#1582 opened Nov 22, 2021 by TimDettmers Loading…
7
Add shared libs
#1567 opened Nov 16, 2021 by RezaYazdaniAminabadi • Draft
Add inference comm
#1552 opened Nov 12, 2021 by RezaYazdaniAminabadi • Draft
1
Refine quantizer for supporting larger hidden-dim and group size
#1544 opened Nov 9, 2021 by RezaYazdaniAminabadi Loading…
Add some improvements for pipeline module, engine and assertion into ds engine
#1529 opened Nov 6, 2021 by hyunwoongko Loading…
modify the quantize.py file for efficiency
#1519 opened Nov 4, 2021 by yaozhewei Loading…
1
parallelize writing of layer checkpoint files across data parallel instances
#1419 opened Sep 30, 2021 by adammoody Loading…
4
Transformer/fix layer norm
#1404 opened Sep 27, 2021 by RezaYazdaniAminabadi Loading…
1
fix dataloder len
#1391 opened Sep 23, 2021 by roywei Loading…
1
9
ProTip! Mix and match filters to narrow down what you’re looking for.