Skip to content

Pull requests: NVIDIA/TransformerEngine

New pull request New
33 Open 827 Closed
33 Open 827 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] Minor optimizations to reduce CPU overheads in modules enhancement New feature or request
#1191 opened Sep 18, 2024 by timmoon10 Loading…
6 of 13 tasks
5
[PyTorch] Fix detection of 3 in 3hd/h3d layouts
#1187 opened Sep 16, 2024 by cyanguwa Loading…
8 of 13 tasks
1
4
Allow to pass architectures like 90a, without being overriden
#1178 opened Sep 11, 2024 by aurianer Loading…
3
[PyTorch] Miscellaneous fixes for FA3 FP8 attention
#1174 opened Sep 10, 2024 by cyanguwa Loading…
9 of 13 tasks
9
[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
[PyTorch] Fused dbias-cast-transpose in bias operation
#1168 opened Sep 6, 2024 by timmoon10 Loading…
7 of 13 tasks
1
Fix autocast deprecation warning.
#1167 opened Sep 6, 2024 by jondeaton Loading…
7
[PyTorch] Activation operations
#1164 opened Sep 6, 2024 by timmoon10 Loading…
6 of 13 tasks
2
[PyTorch/C] Exposed Userbuffers configuration option to control comm and compute stream priorities enhancement New feature or request
#1149 opened Aug 29, 2024 by denera Loading…
8 of 13 tasks
@denera
4
[PyTorch] Avoid saving fp8_tensors in certain scenarios
#1143 opened Aug 28, 2024 by cyanguwa Loading…
8 of 13 tasks
1
[PyTorch] Userbuffers support in operation-based API
#1142 opened Aug 27, 2024 by timmoon10 Loading…
7 of 13 tasks
2
Norms Refractor
#1140 opened Aug 27, 2024 by phu0ngng • Draft
5 of 13 tasks
12
Don't save fp8 q/k/v/out tensors when using bf16 bprop
#1139 opened Aug 27, 2024 by guyueh1 Loading…
13 tasks
1
Fix param input order for cudagraph bug Something isn't working
#1138 opened Aug 27, 2024 by yifeis-nv Loading…
4 of 13 tasks
4
Add high_precision_init_val to model params when using fp8_model_init
#1121 opened Aug 19, 2024 by kunlunl Loading…
8 of 13 tasks
10
[PyTorch] Debug CUDA graph support with operation-based API bug Something isn't working
#1117 opened Aug 16, 2024 by timmoon10 Loading…
7 of 13 tasks
10
Draft: Support using fp16 master weights and fp16/fp8 optimizer states in FusedAdam
#1078 opened Aug 5, 2024 by kunlunl Loading…
7 of 13 tasks
[C/PyTorch] Userbuffers and comm+GEMM overlap algorithms refactored and moved to TE/common enhancement New feature or request
#1067 opened Jul 31, 2024 by denera Loading…
8 of 13 tasks
@denera
26
[PyTorch] Debug checkpointing with operation-based API bug Something isn't working
#1063 opened Jul 31, 2024 by timmoon10 Loading…
8 of 13 tasks
2
Use pyproject.toml to specify build requirements build Build system
#1061 opened Jul 30, 2024 by ksivaman Loading…
6 of 13 tasks
@ksivaman
11
[JAX] Support Ring Attention (Context Parallelism)
#1059 opened Jul 30, 2024 by mingxu1067 • Draft
1 of 13 tasks
Change condition for ub tp overlap.
#1055 opened Jul 29, 2024 by Victarry Loading…
1 of 13 tasks
[PyTorch] Normalization ops enhancement New feature or request
#1033 opened Jul 22, 2024 by timmoon10 Loading…
8 of 13 tasks
31
Flash attention support softcap.
#1013 opened Jul 14, 2024 by Lzhang-hub Loading…
7 tasks
4
[JAX] Sharding Utils
#1003 opened Jul 9, 2024 by mingxu1067 • Draft
8 of 13 tasks
7
ProTip! Add no:assignee to see everything that’s not assigned.