Skip to content

Pull requests: NVIDIA/TransformerEngine

New pull request New
46 Open 918 Closed
46 Open 918 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

[Dummy] Testing branch for #1326 invalid This doesn't seem right
#1330 opened Nov 13, 2024 by timmoon10 • Draft
13 tasks
[PyTorch] Integration test for Megatron-LM bug Something isn't working
#1329 opened Nov 13, 2024 by timmoon10 Loading…
9 of 14 tasks
1
[PyTorch] Fix GQA error message 1.13.0
#1328 opened Nov 12, 2024 by cyanguwa Loading…
8 of 13 tasks
2
[COMMON/JAX] Support sliding window on THD format
#1327 opened Nov 11, 2024 by zlsh80826 Loading…
6 of 13 tasks
6
[PyTorch] Remove special handling for FP8 params in FP8 recipe infrastructure
#1326 opened Nov 9, 2024 by timmoon10 Loading…
8 of 13 tasks
2
Build with uv instead of just pip
#1324 opened Nov 8, 2024 by jennifgcrl Loading…
5 of 13 tasks
1
4
[PyTorch] Fix ONNX export bug with operation-based API bug Something isn't working
#1320 opened Nov 7, 2024 by timmoon10 Loading…
8 of 13 tasks
1
[C] Normalization Refactor + Adding CUDNN backend
#1315 opened Nov 5, 2024 by phu0ngng • Draft
13 tasks
5
TP communication overlap: enable the overlap between GEMM chunk at Ho…
#1311 opened Nov 4, 2024 by erhoo82 Loading…
1 of 13 tasks
Improving communication overlap for the case of multi kernel queue usage
#1308 opened Nov 2, 2024 by youngeunkwon0405 Loading…
13 tasks
11
[JAX] Collective GEMM custom op with nvte_cublas_gemm (no comm. overlap) jax
#1307 opened Nov 2, 2024 by denera • Draft
7 of 17 tasks
@denera
2
[PyTorch] Add heuristics for intializing FP8 params enhancement New feature or request
#1300 opened Oct 30, 2024 by timmoon10 Loading…
8 of 13 tasks
2
Offloading example
#1299 opened Oct 29, 2024 by sanandaraj5597 Loading…
[PyTorch] Fix get_swa_mask() for padding masks
#1281 opened Oct 21, 2024 by cyanguwa Loading…
6 of 13 tasks
4
[PyTorch] Fix autocast deprecation warnings
#1277 opened Oct 21, 2024 by yaox12 Loading…
13 tasks
14
[PyTorch] Remove sequence parallel check for setting dropout RNG context
#1272 opened Oct 18, 2024 by ksivaman • Draft
1 of 13 tasks
attention_mask fill with -inf for UnfusedDotProductAttention
#1268 opened Oct 18, 2024 by Agoniii Loading…
1 of 13 tasks
2
Draft: reduce cudagraph mem via preoallcations
#1253 opened Oct 15, 2024 by JimmyZhang12 Loading…
13 tasks
[pyTorch] Infrastructure for C++ QuantizedTensor
#1251 opened Oct 14, 2024 by ptrendx • Draft
13 tasks
2
fused out correction in CP
#1248 opened Oct 14, 2024 by xiaoyao0115 Loading…
12 tasks
10
Save CUDA Graph memory by reusing input and output tensors
#1234 opened Oct 9, 2024 by buptzyb Loading…
5 of 13 tasks
Support CUDA Graph for MoE models
#1233 opened Oct 9, 2024 by buptzyb Loading…
6 of 13 tasks
23
[PyTorch] Improve CP P2P efficiency
#1208 opened Sep 26, 2024 by yenchenlin Loading…
1 of 6 tasks
Draft: Use fused push_send_recv kernel for TP AG and RS overlaps
#1200 opened Sep 24, 2024 by erhoo82 Loading…
13 tasks
[WIP] [PyTorch] Proof-of-concept for using operation-based API in modules
#1173 opened Sep 10, 2024 by timmoon10 • Draft
2 of 13 tasks
1
ProTip! Add no:assignee to see everything that’s not assigned.