Skip to content

Pull requests: NVIDIA/TransformerEngine

New pull request New
100 Open 1,693 Closed
100 Open 1,693 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
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[JAX] Use TE quant if TE fused act is disabled
#2374 opened Nov 12, 2025 by jberchtold-nvidia Loading…
8 of 13 tasks
4
FSDP2 Allgather Perf improvement and support for FusedAdam with FSDP2
#2370 opened Nov 12, 2025 by vthumbe1503 Loading…
2 of 13 tasks
3
[PyTorch] Enable reference Current Scaling recipe
#2368 opened Nov 11, 2025 by negvet Loading…
13 tasks
5
[PyTorch] Add reset cudagraph interface
#2367 opened Nov 11, 2025 by buptzyb Loading…
1 of 13 tasks
4
[JAX] Improve support and testing for direct recipe usage without autocast contexts
#2366 opened Nov 11, 2025 by jberchtold-nvidia Loading…
8 of 13 tasks
24
[JAX] NVFP4 2D 1x1x for Weight
#2365 opened Nov 10, 2025 by phu0ngng • Draft
13 tasks
[JAX] Shardy rule + QuantizeLayout Rework
#2364 opened Nov 10, 2025 by phu0ngng Loading…
7 of 13 tasks
20
[JAX] cuBlasMp integration for CollectiveGemm custom op
#2361 opened Nov 7, 2025 by denera • Draft
5 of 13 tasks
@denera
Add device-Initiated Grouped GEMM supporting m_splits on device
#2360 opened Nov 7, 2025 by QiZhangNV Loading…
1 of 13 tasks
38
[JAX] Make all jax attention calls use non-packed common calls
#2358 opened Nov 6, 2025 by pggPL Loading…
8 of 13 tasks
5
FA num splits option 2.10.0
#2357 opened Nov 6, 2025 by wdykas Loading…
13 tasks
17
[JAX] Support for checkpointing quantizations
#2356 opened Nov 6, 2025 by jberchtold-nvidia Loading…
8 of 13 tasks
20
[PyTorch] Fix amax computation using output_t data in normalization
#2355 opened Nov 6, 2025 by negvet Loading…
1 of 13 tasks
7
[PyTorch][NVFP4][MOE] NVFP4 Grouped Hadamard Amax Kernel
#2351 opened Nov 6, 2025 by zhongbozhu Loading…
4 of 17 tasks
@zhongbozhu
15
[JAX] NVFP4 scale swizzling via nvte kernel
#2350 opened Nov 5, 2025 by phu0ngng • Draft
13 tasks
More detailed documentation for recipes
#2343 opened Nov 4, 2025 by pggPL • Draft
[Core] Fix inconsistent logic in C++ tensor class
#2330 opened Nov 1, 2025 by timmoon10 Loading…
7 of 13 tasks
6
[Common] Added an optimized gated rowwise MXFP8 SwiGLU kernel
#2328 opened Oct 31, 2025 by Oleg-Goncharov Loading…
5 of 13 tasks
10
[Pytorch] change fused cross entropy backward grad to fp32 and reduce one read/…
#2325 opened Oct 31, 2025 by RandMist Loading…
8 of 13 tasks
3
[Common] Persistent MXFP8 kernel
#2323 opened Oct 30, 2025 by Oleg-Goncharov • Draft
13 tasks
[PyTorch] Implement Selective Activation Checkpointing for LayerNormMLP with checkpoint flag
#2311 opened Oct 28, 2025 by jaimec00 Loading…
7 of 13 tasks
1
107
[JAX] Quickstart documentation
#2310 opened Oct 28, 2025 by tdophung Loading…
6 of 11 tasks
121
[JAX] Make test tolerances stricter
#2306 opened Oct 27, 2025 by jberchtold-nvidia Loading…
8 of 13 tasks
2
Docs fix
#2301 opened Oct 24, 2025 by pggPL Loading…
8 of 12 tasks
54
Fix runtime lib loading logic
#2297 opened Oct 23, 2025 by ksivaman Loading…
8 of 13 tasks
16
ProTip! Updated in the last three days: updated:>2025-11-09.