Skip to content

Pull requests: pytorch/helion

New pull request New
227 Open 3,071 Closed
227 Open 3,071 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
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

[xpu][autodiff][2/N] Avoid implicit CUDA device fallbacks CLA Signed This label is managed by the Meta Open Source bot. module: xpu
#3605 opened Sep 10, 2026 by Stonepia Contributor • Draft
1
[xpu][tests][1/N] Unskip XPU tests CLA Signed This label is managed by the Meta Open Source bot. module: xpu
#3604 opened Sep 10, 2026 by Stonepia Contributor • Draft
1
[metal] Enable autotuning CLA Signed This label is managed by the Meta Open Source bot.
#3603 opened Sep 10, 2026 by aditvenk Contributor Loading…
fix(test) relax BMM autodiff tolerance on A10G CLA Signed This label is managed by the Meta Open Source bot.
#3601 opened Sep 10, 2026 by ethche Contributor Loading…
test_examples.py: enable fp8_gemm for Pallas CLA Signed This label is managed by the Meta Open Source bot.
#3597 opened Sep 9, 2026 by cota Collaborator Loading…
3
[metal] honor begin and step for scalar root grids CLA Signed This label is managed by the Meta Open Source bot.
#3596 opened Sep 9, 2026 by aditvenk Contributor • Draft
2
[Pallas] Benchmark rms_norm-bwd at fp32 on the TPU bridge CLA Signed This label is managed by the Meta Open Source bot.
#3595 opened Sep 8, 2026 by norx1991 Contributor • Draft
[Pallas] Bound each TPU bridge kernel so one slow kernel cannot lose the sweep CLA Signed This label is managed by the Meta Open Source bot.
#3594 opened Sep 8, 2026 by norx1991 Contributor • Draft
Remove redundant barriers for inter-loop RAW hazards
#3582 opened Sep 7, 2026 by NaOHCC • Draft
1
[noland] Hillclimb plan, prompt, and per-step records for CuTe attention on GB300 CLA Signed This label is managed by the Meta Open Source bot.
#3579 opened Sep 7, 2026 by yushangdi Contributor • Draft
[cutedsl] Retune the sm_103 flash seeds onto the one-CTA and resident lowerings CLA Signed This label is managed by the Meta Open Source bot.
#3578 opened Sep 7, 2026 by yushangdi Contributor • Draft
[cutedsl] Gate the causal resident softmax on its schedule, and close two KV-width holes CLA Signed This label is managed by the Meta Open Source bot.
#3577 opened Sep 7, 2026 by yushangdi Contributor Loading…
[cutedsl] Search the KV tile width and seed the sm_103 dense shapes at 160 CLA Signed This label is managed by the Meta Open Source bot.
#3576 opened Sep 7, 2026 by yushangdi Contributor Loading…
[cutedsl] Make the flash KV tile width a real dimension CLA Signed This label is managed by the Meta Open Source bot.
#3575 opened Sep 7, 2026 by yushangdi Contributor Loading…
[cutedsl] Run the dense resident softmax on the one-CTA FA4 pipeline CLA Signed This label is managed by the Meta Open Source bot.
#3574 opened Sep 7, 2026 by yushangdi Contributor Loading…
[cutedsl] Gate the dense flash softmax lowering on its schedule, not a seed match CLA Signed This label is managed by the Meta Open Source bot.
#3573 opened Sep 7, 2026 by yushangdi Contributor Loading…
Warm the GPU to its steady clock before the autotuner's final verification CLA Signed This label is managed by the Meta Open Source bot.
#3572 opened Sep 7, 2026 by yushangdi Contributor • Draft
Add an fa4-tuned attention baseline to the backend comparison CLA Signed This label is managed by the Meta Open Source bot.
#3571 opened Sep 7, 2026 by yushangdi Contributor Loading…
[pallas] read f32 rows in the mask-placement test CLA Signed This label is managed by the Meta Open Source bot.
#3570 opened Sep 6, 2026 by cota Collaborator Loading…
[pallas] name the jagged bmm kernel list in the load-store tests CLA Signed This label is managed by the Meta Open Source bot.
#3569 opened Sep 6, 2026 by cota Collaborator Loading…
[pallas] extract control-flow child-graph walk as a helper CLA Signed This label is managed by the Meta Open Source bot.
#3568 opened Sep 6, 2026 by cota Collaborator Loading…
[pallas] reject ordered carry if pallas_loop_type=unroll CLA Signed This label is managed by the Meta Open Source bot.
#3567 opened Sep 6, 2026 by cota Collaborator Loading…
[Pallas] Share nested DMA address setup across streams CLA Signed This label is managed by the Meta Open Source bot.
#3566 opened Sep 5, 2026 by AmesingFlank Contributor • Draft
[Pallas] Pipeline nested DMA panels across outer loops CLA Signed This label is managed by the Meta Open Source bot.
#3565 opened Sep 5, 2026 by AmesingFlank Contributor • Draft
Ellipsis indexing for StackTensorType CLA Signed This label is managed by the Meta Open Source bot.
#3563 opened Sep 4, 2026 by parsshar-RH Contributor Loading…
ProTip! Updated in the last three days: updated:>2026-09-07.