-
Notifications
You must be signed in to change notification settings - Fork 160
Pull requests: pytorch/helion
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[cute] Omit redundant persistent shared loop
CLA Signed
This label is managed by the Meta Open Source bot.
helion-rag: four-arm head-to-head autotuning experiment (LFBO / LLM / hybrid / RAG-LLM)
CLA Signed
This label is managed by the Meta Open Source bot.
#3292
opened Aug 5, 2026 by
IshanAryendu
Contributor
•
Draft
[examples] Add FlashKDA as a second baseline for kda_varlen
CLA Signed
This label is managed by the Meta Open Source bot.
[examples] Read only the causal columns in chunk_fwd_A_diag_anchored
CLA Signed
This label is managed by the Meta Open Source bot.
[examples] Add a variable-length KDA forward
CLA Signed
This label is managed by the Meta Open Source bot.
[examples] Apply the KDA input preamble in the kernels
CLA Signed
This label is managed by the Meta Open Source bot.
Fix BackendImplementationMissing: repair codegen registration skipped by circular-import partial cache
CLA Signed
This label is managed by the Meta Open Source bot.
meta-exported
#3286
opened Aug 5, 2026 by
mengluy0125
Contributor
Loading…
[Pallas] Add eager autotune baselines for rms_norm on TPU
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Sample TPU bridge shapes equally-spaced (cap kl_div / rms_norm to feasible shapes)
CLA Signed
This label is managed by the Meta Open Source bot.
#3284
opened Aug 4, 2026 by
norx1991
Contributor
Loading…
Fix _find_device host overhead in tuple/dict recursion
CLA Signed
This label is managed by the Meta Open Source bot.
meta-exported
#3283
opened Aug 4, 2026 by
mengluy0125
Contributor
Loading…
Enabling FlyDSL in Helion with support for softmax and rms_norm kernels.
CLA Signed
This label is managed by the Meta Open Source bot.
#3281
opened Aug 4, 2026 by
umechand-amd
Collaborator
•
Draft
[Pallas] compact worklist: never request more scoped VMEM than the device allows
CLA Signed
This label is managed by the Meta Open Source bot.
#3276
opened Aug 4, 2026 by
ethche
Contributor
Loading…
torch_tpu: update pin to 1cdd6f0a799a34c12576f2cf29cd242067943814
CLA Signed
This label is managed by the Meta Open Source bot.
[pallas] compose narrowing views on resident VMEM refs
CLA Signed
This label is managed by the Meta Open Source bot.
#3273
opened Aug 4, 2026 by
ethche
Contributor
Loading…
[Pallas] Add SparseCore BF16 lane packing
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Validate composable SparseCore kernels
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Integrate SparseCore as a compiler target
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Build SparseCore launcher transforms
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Render scheduled SparseCore pipelines
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Compose SparseCore FP32 computation
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Schedule SparseCore programs from dependencies
CLA Signed
This label is managed by the Meta Open Source bot.
[Pallas] Build SparseCore memory plans independently
CLA Signed
This label is managed by the Meta Open Source bot.
[Compiler] Make tiled execution strategies extensible
CLA Signed
This label is managed by the Meta Open Source bot.
[NPU][autotuner] config-layer tolerance for Ascend NPU
CLA Signed
This label is managed by the Meta Open Source bot.
#3234
opened Aug 3, 2026 by
yzchang-plus
Loading…
Previous Next
ProTip!
Adding no:label will show everything without a label.