Skip to content

Pull requests: NVIDIA/apex

New pull request New
70 Open 569 Closed
70 Open 569 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 Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

fix building torch extension with glog
#1791 opened Mar 31, 2024 by petronny Loading…
1
Add xentropy bf16 support
#1790 opened Mar 29, 2024 by zyeric Loading…
64-bit indexing Adam
#1786 opened Mar 21, 2024 by cdm114514 Loading…
[CUDNN][cudnn-frontend] Bump cuDNN to 1.0.3
#1777 opened Feb 9, 2024 by eqy Loading…
1
Use master weights for bfloat16 FusedAdam when master_weights=True
#1731 opened Sep 22, 2023 by cbcase Loading…
3
Print the TORCH_CUDA_ARCH_LIST
#1705 opened Aug 7, 2023 by yncxcw Loading…
1
Make distributed fused lamb test names friendly to keyword filtering contrib
#1698 opened Jul 20, 2023 by crcrpar Loading…
Add custom build backend to support build args
#1692 opened Jul 6, 2023 by janEbert Loading…
6
A FasterRMSNorm implementation (based on FasterLayerNorm)
#1688 opened Jun 30, 2023 by Njuapp Loading…
1
Updating missing build dependency in pyproject.toml
#1680 opened Jun 13, 2023 by loadams Loading…
5
Calendar Versioning
#1648 opened Apr 21, 2023 by crcrpar • Draft
CUDA_HOME not checked before use
#1633 opened Apr 5, 2023 by x0152 Loading…
1
[sparse]update support for arbitrary N:M settings sparse
#1631 opened Apr 2, 2023 by LeiWang1999 Loading…
6
FusedAdam as a drop-in replacement for AdamW
#1615 opened Mar 14, 2023 by thomasw21 Loading…
fix bug when CUDA_HOME is None
#1610 opened Mar 13, 2023 by xiawu Loading…
3
Hack to load old distributed optimizer checkpoints
#1602 opened Mar 2, 2023 by timmoon10 • Draft 23.04
Remove deprecated features
#1589 opened Feb 15, 2023 by crcrpar • Draft
1
Fix incorrect initialization of GenericFusedScaleMaskSoftmax
#1586 opened Feb 15, 2023 by qmdnls Loading…
Instance norm nvfuser
#1582 opened Feb 9, 2023 by crcrpar • Draft
1
Fix FusedAdam.zero_grad(set_to_none=True)
#1579 opened Feb 6, 2023 by NouamaneTazi Loading…
4
Fix typo in fp16util.py
#1576 opened Feb 4, 2023 by eltociear Loading…
Support grad_output with noncontiguous strides in LinearWithGradAccum…
#1573 opened Jan 23, 2023 by cat-state Loading…
1
Optimize halo exchange in spatial-parallel bottleneck layer
#1569 opened Jan 20, 2023 by timmoon10 Loading…
Instance norm (and batch, layer norm) using NVFuser Python frontend only
#1564 opened Jan 17, 2023 by jacobhinkle • Draft
27
Layernorm one_plus support in fast_layer_norm
#1557 opened Dec 12, 2022 by anmolgupt Loading…
3
ProTip! Add no:assignee to see everything that’s not assigned.