Skip to content

Pull requests: pggPL/TransformerEngine

New pull requestNew
23 Open 15 Closed
23 Open 15 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 reviewsReview requiredApproved reviewChanges requested
Assignee
Filter by who’s assigned
Assigned to nobodyLoading
Sort
Sort by
NewestOldestMost commentedLeast commentedRecently updatedLeast recently updatedBest match
Most reactions
👍👎😄🎉😕❤️🚀👀

Pull requests list

[PyTorch] Schedule delayed-scaling updates after backward
#39 opened Aug 20, 2026 by pggPLOwnerLoading…
9 of 13 tasks
[PyTorch] Delegate delayed-scaling updates to RecipeState
#38 opened Aug 20, 2026 by pggPLOwnerLoading…
8 of 13 tasks
Loosen fp16 atol in ONNX export tests for onnxruntime 1.29 CPU numerics
#37 opened Aug 20, 2026 by pggPLOwnerLoading…
7 of 13 tasks
[PyTorch] DeepSeekV3Layer: full MoE transformer layer (MLA + DeepSeek MoE)
#36 opened Aug 18, 2026 by pggPLOwner • Draft
7 of 13 tasks
CI: minimum supported PyTorch check
#35 opened Aug 18, 2026 by pggPLOwnerLoading…
[PyTorch] Finalize quantization updates at logical backward boundaries
#34 opened Aug 17, 2026 by pggPLOwner • Draft
7 of 8 tasks
[PyTorch] [torch.compile] torch.compile support for GroupedLinear
#32 opened Aug 13, 2026 by pggPLOwner • Draft
7 of 13 tasks
[WIP] Torch stable ABI foundation: torch_compat layer + stable Tensor pybind caster
#31 opened Aug 10, 2026 by pggPLOwner • Draft
7 of 13 tasks
[PyTorch] Declare Bias and the activations as custom ops
#30 opened Aug 5, 2026 by pggPLOwner • 2/2Loading…
7 of 13 tasks
[PyTorch] torch.compile for an OperationFuser group holding one operation
#29 opened Aug 5, 2026 by pggPLOwner • 1/2Loading…
7 of 13 tasks
Carry symbolic shapes through compile custom ops community-contribution
#26 opened Aug 4, 2026 by shino16Loading…
13 tasks
[PyTorch][CI] Fix test issues: connect test files that are not wired into qa
#25 opened Jul 30, 2026 by pggPLOwnerLoading…
8 of 13 tasks
[JAX] Fix typo in qkv_layout docstring: seperate -> separate ai-approved Reviewed by ai_pr_bot and judged trivially simple
#23 opened Jul 27, 2026 by pggPLOwnerLoading…
[PyTorch][torch.compile] Support for DotProductAttention on flash and unfused backends
#22 opened Jul 27, 2026 by pggPLOwnerLoading…
9 tasks done
[PyTorch] Centralize TE↔torch binary boundary in a single facade file
#21 opened Jul 17, 2026 by pggPLOwnerLoading…
5 of 6 tasks
[PyTorch] torch.compile support for UnfusedDotProductAttention
#19 opened Jul 8, 2026 by pggPLOwnerLoading…
7 of 13 tasks
[PyTorch] DotProductAttention: declarative packed qkv/kv inputs
#18 opened Jul 7, 2026 by pggPLOwnerLoading…
[common/PyTorch] Fused attention: pass real tensor strides to cuDNN (opt-in, v2 C API)
#17 opened Jul 7, 2026 by pggPLOwnerLoading…
[PyTorch] Compile-friendly attention API: pointer-free qkv layout detection
#16 opened Jul 7, 2026 by pggPLOwnerLoading…
[PyTorch] Deprecate internal max_seqlen derivation in DotProductAttention
#15 opened Jul 3, 2026 by pggPLOwnerLoading…
[PyTorch][torch.compile] Replace TensorProto with make_empty_traceable community-contribution
#13 opened Jul 2, 2026 by kshitij12345Loading…
13 tasks
6
[PyTorch][torch.compile] Add TensorProto mechanism
#8 opened Jun 6, 2026 by pggPLOwnerLoading…
4 of 13 tasks
28
[PyTorch][torch.compile] Make quantizers opaque value objects
#7 opened Jun 6, 2026 by pggPLOwnerLoading…
8 of 13 tasks
3
ProTip! Filter pull requests by the default branch with base:main.