Skip to content

Pull requests: pytorch/executorch

New pull request New
438 Open 17,562 Closed
438 Open 17,562 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

Gemma4 31b tq4 prefill decode tuned CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20512 opened Jun 25, 2026 by Gasoonjia Contributor • Draft
1
Disable qnn_16a16w Llama runner test (OOM on linux.2xlarge) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20511 opened Jun 25, 2026 by psiddh Contributor Loading…
8
[ExecuTorch][WebGPU] SDPA: branchless aligned/tail loads in the QK/AV kernels CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20510 opened Jun 25, 2026 by pytorchbot Collaborator Loading…
1
[ExecuTorch][WebGPU] SDPA: skip QK contraction for fully-masked causal tiles CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20509 opened Jun 25, 2026 by pytorchbot Collaborator Loading…
1
[ExecuTorch][WebGPU] Coalesce SDPA AV V-cache reads along contiguous head-dim CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20508 opened Jun 25, 2026 by pytorchbot Collaborator Loading…
1
[ExecuTorch][WebGPU] Register-tile the SDPA QK/AV kernels CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20507 opened Jun 25, 2026 by pytorchbot Collaborator Loading…
2
[gemma4_31b][cuda] length-aware bf16 global attention + head_dim-agno… CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20506 opened Jun 25, 2026 by Gasoonjia Contributor • Draft
2
Add WebGPU sigmoid operator CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. release notes: ops & kernels Changes to the opset and any new / changed kernel implementations
#20504 opened Jun 25, 2026 by iamorlando • Draft
4
Add adapter_quant field to LoraConfig CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20503 opened Jun 25, 2026 by billmguo Contributor Loading…
4
Inline per-tensor SIMD fast path in fusion_g3 op_dequantize (recreate D108798741) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20499 opened Jun 24, 2026 by zonglinpeng Contributor Loading…
4
Use caller CUDA stream for D2H and H2D copies (#20498) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20498 opened Jun 24, 2026 by Conarnar Contributor Loading…
4
Fix buck dep for test_permute_optimization_passes (#20497) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20497 opened Jun 24, 2026 by digantdesai Contributor Loading…
4
Add oncall to executorch/backends/aoti/TARGETS CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20496 opened Jun 24, 2026 by Ben0mega Contributor Loading…
4
Enable buck-native x86 simulator test for QNN op tests (#20494) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20494 opened Jun 24, 2026 by billmguo Contributor Loading…
4
[ET-VK][quantized] Store dq8ca per-token zero-point as fp32 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. meta-exported
#20491 opened Jun 24, 2026 by SS-JIA Contributor Loading…
3
Qualcomm AI Engine Direct - LLM multi-batch quantization and evaluation CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. release notes: qualcomm Changes to the Qualcomm backend delegate
#20488 opened Jun 24, 2026 by DannyYuyang-quic Contributor Loading…
3
[executorch][cuda] fuse gate/up MLP projections CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20482 opened Jun 24, 2026 by Gasoonjia Contributor Loading…
3
[executorch][gemma4] fuse MLP gate/up at GGUF load CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20481 opened Jun 24, 2026 by Gasoonjia Contributor • Draft
2
[gemma4_31b][cuda] Export Gemma4-31B @128k on 5090 CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20480 opened Jun 24, 2026 by Gasoonjia Contributor Loading…
5
Arm backend: Add TOSA binary op visitors ciflow/trunk CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. module: arm Issues related to arm backend partner: arm For backend delegation, kernels, demo, etc. from the 3rd-party partner, Arm release notes: arm Changes to the ARM backend delegate
#20479 opened Jun 24, 2026 by SaoirseARM Collaborator Loading…
6
Qualcomm AI Engine Direct - Testing fix CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20478 opened Jun 24, 2026 by winskuo-quic Collaborator • Draft
2
gemma4_31b: add OpenAI serving entrypoint ciflow/cuda CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20473 opened Jun 24, 2026 by mergennachin Contributor Loading…
14
Qualcomm: cap inf replacement value to fix 16a16w accuracy regression CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20471 opened Jun 24, 2026 by psiddh Contributor Loading…
14
[ExecuTorch][WebGPU] aten.index.Tensor test suite (export + native golden) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20465 opened Jun 23, 2026 by JulianCloudNTH Contributor Loading…
2
[ExecuTorch][WebGPU] Add aten.index.Tensor (1D-self gather) CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
#20464 opened Jun 23, 2026 by JulianCloudNTH Contributor Loading…
2
ProTip! What’s not been updated in a month: updated:<2026-05-25.