Search results

Conv2d with padding 'same' fails (when quantised) at runtime on XNNPACK

Status: Open.

#20517 In pytorch/executorch;

· m-lawre opened on Jun 25, 2026

ARM Ethos-U: Operators not properly quantized when cat is present

module: armIssues related to arm backendIssues related to arm backend

partner: armFor backend delegation, kernels, demo, etc. from the 3rd-party partner, ArmFor backend delegation, kernels, demo, etc. from the 3rd-party partner, Arm

Status: Open.

Bug

#20486 In pytorch/executorch;

· etrommer opened on Jun 24, 2026

Windows on ARM64 (win_arm64) support — plans, and torch 2.13 bump?

Status: Open.

#20476 In pytorch/executorch;

· vortex-captain opened on Jun 24, 2026

Bump C++ standard to C++20

rfcRequest for comment and feedback on a post, proposal, etc.Request for comment and feedback on a post, proposal, etc.

Status: Open.

#20468 In pytorch/executorch;

· rascani opened on Jun 23, 2026

ETDump adds runtime overhead by serializing the flatbuffer during inference

Status: Open.

#20467 In pytorch/executorch;

· rascani opened on Jun 23, 2026

[RFC] ExecuTorch KV cache management

Status: Open.

#20431 In pytorch/executorch;

· metascroy opened on Jun 22, 2026

XNNPACK Buck build appears not to enable KleidiAI/SME2 while CMake enables KleidiAI by default

module: xnnpackIssues related to xnnpack delegation and the code under backends/xnnpack/Issues related to xnnpack delegation and the code under backends/xnnpack/

partner: armFor backend delegation, kernels, demo, etc. from the 3rd-party partner, ArmFor backend delegation, kernels, demo, etc. from the 3rd-party partner, Arm

Status: Open.

#20378 In pytorch/executorch;

· mansnils opened on Jun 18, 2026

# Good First Issue: In-Model Sampling Head for the MLX Backend (Gumbel Sampling)

good first issueGood for newcomersGood for newcomers

Status: Open.

#20353 In pytorch/executorch;

· metascroy opened on Jun 17, 2026

[MLX] Good First Issue: Skip MoE expert sorting during decode (MLX backend, Qwen 3.5 MoE)

good first issueGood for newcomersGood for newcomers

Status: Open.

#20352 In pytorch/executorch;

· metascroy opened on Jun 17, 2026

Support CUDA graph replay with multi-session LLM serving

Status: Open.

#20310 In pytorch/executorch;

· mergennachin opened on Jun 16, 2026

Exynos 2400 Support

module: samsung

partner: samsungFor backend delegation, kernels, demo, etc. from the 3rd-party partner, SamsungFor backend delegation, kernels, demo, etc. from the 3rd-party partner, Samsung

Status: Open.

Feature

#20303 In pytorch/executorch;

· harryjulian opened on Jun 16, 2026

kernel 'aten::bucketize.Tensor_out' not found.

module: androidIssues related to Android code, build, and executionIssues related to Android code, build, and execution

module: kernelsIssues related to kernel libraries and utilities, and code under kernels/Issues related to kernel libraries and utilities, and code under kernels/

Status: Open.

Task

#20270 In pytorch/executorch;

· Gallinator opened on Jun 14, 2026