Issues

• CUDA: LeRobot policies that do not run yet, and what blocks them

  module: cuda
  pytorch/executorch#24684 · Filter by author shoumikhin opened on Nov 5, 2026

• CUDA AOTI: Inductor memory planning fails with undefined symbol aoti_torch__alloc_from_pool

  module: cuda
  pytorch/executorch#24682 · Filter by author mshr-h opened on Nov 5, 2026
  1

  

• [CPU] Shared elementwise helpers ignore output strides during broadcasting

  pytorch/executorch#24671 · Filter by author mshr-h opened on Nov 5, 2026

• Core ML and MLX: LeRobot policies that do not run yet, and what blocks them

  module: coremlmodule: mlx
  pytorch/executorch#24669 · Filter by author shoumikhin opened on Nov 5, 2026

• Qualcomm (QNN): LeRobot policies that do not run yet, and what blocks them

  module: qnn
  pytorch/executorch#24668 · Filter by author shoumikhin opened on Nov 5, 2026
  1

• XNNPACK: LeRobot policies that do not run yet, and what blocks them

  module: xnnpack
  pytorch/executorch#24667 · Filter by author shoumikhin opened on Nov 5, 2026
  1

• Qualcomm: bmm over a flattened 4D input returns ~1/W of the right value when followed by unsqueeze (HTP fp16)

  pytorch/executorch#24502 · Filter by author msluszniak opened on Nov 3, 2026

• Optimized le.Tensor_out returns b <= a for reverse-argument broadcasts

  pytorch/executorch#24485 · Filter by author SofianElmotiem opened on Nov 3, 2026

• Qualcomm AI Engine Direct - fp16a8w quantization crashes on layer_norm

  pytorch/executorch#24434 · Filter by author msluszniak opened on Nov 2, 2026

• install_executorch.sh fails in a fresh uv virtual environment: No module named pip

  pytorch/executorch#24417 · Filter by author mshr-h opened on Nov 2, 2026

• [aarch64/GCC] CPU kernel tests fail to compile due to Half narrowing conversions

  pytorch/executorch#24410 · Filter by author mshr-h opened on Nov 2, 2026

• [CPU] Complex scalar broadcasting fails or returns incorrect results

  pytorch/executorch#24408 · Filter by author mshr-h opened on Nov 2, 2026
  3

• TensorRT: LeRobot policies that do not run yet, and what blocks them

  module: cuda
  pytorch/executorch#24353 · Filter by author shoumikhin opened on Oct 30, 2026

• More Exynos SoCs support

  pytorch/executorch#24350 · Filter by author amirali1390jeddi opened on Oct 30, 2026

• torch.complex fails Edge export and CPU execution

  bugmodule: kernelstriaged
  Bug

  pytorch/executorch#24328 · Filter by author mshr-h opened on Oct 30, 2026
  

• OpenVINO: LeRobot policies that do not run yet, and what blocks them

  module: openvino
  pytorch/executorch#24325 · Filter by author shoumikhin opened on Oct 30, 2026

• [Rpi5 setup] Make Raspberry Pi cross-compiled binaries run out of the box on supported OS versions

  module: microcontrollersrfc
  Task

  pytorch/executorch#24228 · Filter by author BicMak opened on Oct 28, 2026
  1

  

• [ET-VK] fp16 linear on an RDNA3.5 iGPU: default (texture) export runs at ~1.1 TFLOP/s, coopmat unreachable without buffer storage, and the stock buffer path is ~20x less accurate

  module: vulkanrfc
  Task

  pytorch/executorch#24196 · Filter by author giuliocorradi opened on Oct 27, 2026
  1

  

• [QNN] 16a8w: gathered tensor read wrong by a consumer that only feeds a graph output

  bugmodule: qnnpartner: qualcomm
  Bug

  pytorch/executorch#24125 · Filter by author msluszniak opened on Oct 23, 2026
  

• Elastic on-device inference - swap LLM mid-generation and carry KV cache (molt / kvine)

  module: llm/evaluationrfc
  Feature

  pytorch/executorch#24106 · Filter by author NagaYu opened on Oct 23, 2026
  

• Arm/TOSA: allow int64 through when the spec declares the int64 extension

  bugmodule: armpartner: arm
  Bug

  pytorch/executorch#24095 · Filter by author abm-77 opened on Oct 22, 2026
  4

  

• executor_runner segfaults in portable kernels when a .pte tensor has neither a data buffer nor a memory plan (null data pointer)

  buggood first issuemodule: runtimetriaged
  Bug

  pytorch/executorch#23955 · Filter by author professor-moody opened on Oct 18, 2026
  1

  

• [QNN] 16a4w_block stride-2 conv + ReLU fails HTP graph prepare on SM8850 (Relu16bParamsPlaceholder)

  bugmodule: qnnpartner: qualcomm
  Bug

  pytorch/executorch#23922 · Filter by author msluszniak opened on Oct 16, 2026
  1

  

• Cortex-M backend replaces channels-first dw conv with channels-last dw conv without handling the changed layout, fails at runtime

  bugmodule: microcontrollerspartner: arm
  pytorch/executorch#23812 · Filter by author Erik-Lundell opened on Oct 14, 2026
  2

• macOS Python wheel _portable_lib has non-relocatable libtorch rpath

  pytorch/executorch#23742 · Filter by author robell opened on Oct 13, 2026
  1