Skip to content

Pull requests: NVIDIA/TransformerEngine

New pull request New
64 Open 1,109 Closed
64 Open 1,109 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

fix a sync race error of softmax_lse in CP+THD+P2P
#1624 opened Mar 29, 2025 by xrennvidia Loading…
5 of 13 tasks
1
[PyTorch] Support default process group with FP8 current scaling bug Something isn't working
#1621 opened Mar 27, 2025 by timmoon10 Loading…
6 of 13 tasks
2
[PyTorch] Debug NCCL communication overlapping in linear backward with FP8 data bug Something isn't working
#1620 opened Mar 27, 2025 by timmoon10 Loading…
5 of 13 tasks
1
[PyTorch] Make breaking change in InferenceParams.init more explicit 2.2.0
#1619 opened Mar 26, 2025 by cyanguwa Loading…
8 of 13 tasks
2
[PyTorch] fix fuse_wgrad_accumulation in LayerNormMLP backward
#1618 opened Mar 26, 2025 by Marks101 Loading…
1 of 6 tasks
4
Tongliu fp8 a2a
#1617 opened Mar 26, 2025 by Autumn1998 Loading…
13 tasks
[Pytorch] NVIDIA-DL-Framework-Inspect support – part 1 – core
#1614 opened Mar 25, 2025 by pggPL Loading…
7 tasks done
1
[Pytorch] NVIDIA-DL-Framework-Inspect support – part 2 – features
#1613 opened Mar 25, 2025 by pggPL Loading…
7 tasks done
[Pytorch] NVIDIA-DL-Framework-Inspect support – part 3 – tests
#1612 opened Mar 25, 2025 by pggPL Loading…
7 of 13 tasks
[Pytorch] NVIDIA-DL-Framework-Inspect support – part 4 – documentation
#1611 opened Mar 25, 2025 by pggPL Loading…
7 tasks done
Improved performance of mxfp8 cast kernels 2.2.0 performance Performance issues
#1602 opened Mar 22, 2025 by Oleg-Goncharov Loading…
6 of 13 tasks
10
[JAX] Add fast path for causal masking with segment IDs.
#1601 opened Mar 21, 2025 by mgoldfarb-nvidia Loading…
8 of 13 tasks
6
[PyTorch] Tutorial for the ONNX export
#1586 opened Mar 18, 2025 by pggPL Loading…
8 of 13 tasks
[JAX] Unbalanced Context Parallelism with THD format
#1565 opened Mar 12, 2025 by zlsh80826 Loading…
8 of 13 tasks
3
Draft: split wgrad for GroupedLinear
#1564 opened Mar 12, 2025 by lhb8125 • Draft
13 tasks
2
[CI] Add isort
#1563 opened Mar 12, 2025 by yaox12 • Draft
13 tasks
Enable AttnFuncWithCPAndKVP2P to support mla
#1561 opened Mar 12, 2025 by SuperCB Loading…
3 of 13 tasks
Blockwise scaling linear quantization recipe
#1559 opened Mar 11, 2025 by kwyss-nvidia Loading…
8 of 13 tasks
8
change softmax_lse correction of CP to FP32
#1546 opened Mar 7, 2025 by xrennvidia Loading…
6 of 13 tasks
1
Subchannel Block quantized GEMM
#1545 opened Mar 6, 2025 by kwyss-nvidia Loading…
6 of 12 tasks
1
Fused Linear and Cross Entropy operations
#1537 opened Mar 5, 2025 by Jianbing-D Loading…
2
[MoE] Enable MXFP8 and Per-Tensor Current Scaling for Grouped Linear
#1525 opened Feb 28, 2025 by yaox12 Loading…
5 of 17 tasks
17
Blockwise float8 quantizer and quantized tensor class
#1513 opened Feb 27, 2025 by kwyss-nvidia Loading…
23 of 34 tasks
26
Draft: split wgrad poc
#1510 opened Feb 26, 2025 by lhb8125 • Draft
13 tasks
[Pytorch] Dynamo ONNX export support
#1497 opened Feb 19, 2025 by pggPL Loading…
8 of 13 tasks
17
ProTip! Filter pull requests by the default branch with base:main.