Skip to content

Pull requests: NVIDIA/TransformerEngine

New pull request New
32 Open 704 Closed
32 Open 704 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
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

[pyTorch] Fix wrong results for noncontiguous input
#1017 opened Jul 16, 2024 by ptrendx Loading…
8 of 13 tasks
3
[PyTorch] Custom kernel to compute reciprocal of a single float enhancement New feature or request
#1016 opened Jul 15, 2024 by timmoon10 • Draft
5 of 13 tasks
5
Flash attention support softcap.
#1013 opened Jul 14, 2024 by Lzhang-hub Loading…
7 tasks
1
Fix context parallelism implementation with THD format
#1012 opened Jul 13, 2024 by xrennvidia Loading…
6 of 13 tasks
Initialize output tensors to 0 for THD (temporary)
#1009 opened Jul 12, 2024 by cyanguwa Loading…
8 of 13 tasks
6
[JAX] Allow enabling partial custom calls through the environment variable
#1007 opened Jul 10, 2024 by zlsh80826 Loading…
8 of 13 tasks
15
DGRAD_RS UB overlap Bug fixes
#1004 opened Jul 10, 2024 by vasunvidia Loading…
13 tasks
2
[JAX] Sharding Utils
#1003 opened Jul 9, 2024 by mingxu1067 • Draft
8 of 13 tasks
7
Optimize multi-tensor cast-transpose kernel enhancement New feature or request
#998 opened Jul 8, 2024 by timmoon10 • Draft
6 of 13 tasks
2
Add efficient cross entropy by cuda kernel.
#995 opened Jul 8, 2024 by cb521 Loading…
13 tasks
[PyTorch] Fixing hang in initialize_ub() for multi-node runs after PR901 removal of MPI-dependence bug Something isn't working
#986 opened Jul 3, 2024 by denera Loading…
8 of 13 tasks
@denera
3
[pre-commit.ci] pre-commit suggestions wontfix This will not be worked on
#979 opened Jul 2, 2024 by pre-commit-ci bot • Draft
2
[PyTorch] Support dtype casting in fused adam
#977 opened Jul 1, 2024 by Wong4j Loading…
8 of 13 tasks
30
[Paddle] Add deterministic option in DotProductAttention
#956 opened Jun 23, 2024 by Wong4j Loading…
8 of 13 tasks
15
Lower memory usage during AttnFuncWithCP.forward
#951 opened Jun 21, 2024 by i4never Loading…
8 of 13 tasks
1
2
[TE/JAX] Prototype for New XLA Custom Calls with FFI enhancement New feature or request jax
#946 opened Jun 19, 2024 by phu0ngng Loading…
3 of 13 tasks
12
[PyTorch] Add option to pass kwargs to CUDA graph module enhancement New feature or request
#945 opened Jun 19, 2024 by timmoon10 Loading…
9 of 13 tasks
4
Expose rotary_base as an arg instead of hardcoding
#944 opened Jun 18, 2024 by sudhakarsingh27 Loading…
1 of 6 tasks
5
[MoE][Common/PyTorch] Add permutation enhancement New feature or request
#936 opened Jun 17, 2024 by StudyingShao Loading…
13 tasks
26
Fp8 model init factory
#880 opened May 30, 2024 by sudhakarsingh27 • Draft
3
Avoid framework specific import from top level enhancement New feature or request
#862 opened May 22, 2024 by ksivaman • Draft
6 of 11 tasks
@ksivaman
1
Generation tutorial for Gemma model
#829 opened May 1, 2024 by pggPL Loading…
8 of 11 tasks
34
[Pytorch] Implement fp32 accumulation for attention with context parallel in both forward and backward pass.
#821 opened Apr 28, 2024 by Yuxin-CV Loading…
1
[UB] Adding support for multinode nvlink
#815 opened Apr 26, 2024 by shamisp Loading…
11
Bug fix in DGRAD->RS overlap
#802 opened Apr 23, 2024 by vasunvidia • Draft
ProTip! Find all pull requests that aren't related to any open issues with -linked:issue.