Search results

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

Status: Open.

#20303 In pytorch/executorch;

· harryjulian opened on Jun 16, 2026

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

Status: Open.

#20270 In pytorch/executorch;

· Gallinator opened on Jun 14, 2026

URML (open robot intent language): a validated-action gate above on-device policy inference (request for comment)

Status: Open.

#20268 In pytorch/executorch;

· idoco2003 opened on Jun 13, 2026

Vulkan: elided copy ops (lift_fresh_copy/clone) corrupt sibling outputs (memory-planner aliasing)

module: vulkanIssues related to the Vulkan delegate and code under backends/vulkan/Issues related to the Vulkan delegate and code under backends/vulkan/

Status: Open.

#20257 In pytorch/executorch;

· elliott-wen opened on Jun 12, 2026

OOM with kokoro models

Status: Open.

#20178 In pytorch/executorch;

· maifeeulasad opened on Jun 10, 2026

Qwen3-0.6B: QNN HTP produces degenerate output at all precisions (XNNPACK correct on same device)

module: qnnIssues related to Qualcomm's QNN delegate and code under backends/qualcomm/Issues related to Qualcomm's QNN delegate and code under backends/qualcomm/

Status: Open.

Task

#20168 In pytorch/executorch;

· psiddh opened on Jun 9, 2026

[ARM] EthosUQuantizer QAT prepare_qat_pt2e ignores model parameter (optimizer got an empty parameter list)

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.

#20148 In pytorch/executorch;

· jonasdaugalas opened on Jun 9, 2026

Linux and Windows support for Vulkan

module: vulkanIssues related to the Vulkan delegate and code under backends/vulkan/Issues related to the Vulkan delegate and code under backends/vulkan/

Status: Open.

Feature

#20140 In pytorch/executorch;

· Reubend opened on Jun 9, 2026