forked from NVIDIA/TransformerEngine
-
Notifications
You must be signed in to change notification settings - Fork 1
Pull requests: pggPL/TransformerEngine
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[PyTorch] torch.compile support for UnfusedDotProductAttention
#19
opened Jul 8, 2026 by
pggPL
Owner
Loading…
7 of 13 tasks
[PyTorch] DotProductAttention: declarative packed qkv/kv inputs
#18
opened Jul 7, 2026 by
pggPL
Owner
Loading…
[common/PyTorch] Fused attention: pass real tensor strides to cuDNN (opt-in, v2 C API)
#17
opened Jul 7, 2026 by
pggPL
Owner
Loading…
[PyTorch] Compile-friendly attention API: pointer-free qkv layout detection
#16
opened Jul 7, 2026 by
pggPL
Owner
Loading…
[PyTorch] Deprecate internal max_seqlen derivation in DotProductAttention
#15
opened Jul 3, 2026 by
pggPL
Owner
Loading…
[PyTorch][torch.compile] Replace TensorProto with make_empty_traceable
community-contribution
#13
opened Jul 2, 2026 by
kshitij12345
Loading…
13 tasks
[PyTorch] Add torch.compile custom-op path for Linear
#9
opened Jun 7, 2026 by
pggPL
Owner
Loading…
13 tasks
[PyTorch][torch.compile] Add TensorProto mechanism
#8
opened Jun 6, 2026 by
pggPL
Owner
Loading…
4 of 13 tasks
[PyTorch][torch.compile] Make quantizers opaque value objects
#7
opened Jun 6, 2026 by
pggPL
Owner
Loading…
8 of 13 tasks
ProTip!
Find all pull requests that aren't related to any open issues with -linked:issue.