Search results

<code>[__recompiles] - 0/3: expected type of 'args[1]' to be a tensor type, ' but found &lt;class 'torch.Tensor'&gt;</code> cryptic recompilation cause

module: dynamo

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#165051 In pytorch/pytorch;

· fxmarty-amd opened on Oct 9, 2025

Official support for sm_120 (RTX 50-series / Blackwell) in stable PyTorch builds

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: windowsWindows support for PyTorchWindows support for PyTorch

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#164342 In pytorch/pytorch;

· endvntgf-design opened on Oct 1, 2025

Internal assert failed when using Tensorly on MPS

module: mpsRelated to Apple Metal Performance Shaders frameworkRelated to Apple Metal Performance Shaders framework

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163962 In pytorch/pytorch;

· ivegner opened on Sep 26, 2025

ModuleNotFoundError: No module named 'importlib_metadata'

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

oncall: distributedAdd this issue/PR to distributed oncall triage queueAdd this issue/PR to distributed oncall triage queue

Status: Open.

#163946 In pytorch/pytorch;

· yunyiyun opened on Sep 26, 2025

Enabling activation checkpoint debug makes throughput gradually slow down to crawl.

module: activation checkpointingRelated to activation checkpointingRelated to activation checkpointing

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163923 In pytorch/pytorch;

· githubsgi opened on Sep 26, 2025

building pytorch c++ extension with cuda 13 causes unexpected behavior when pytorch is built with cuda 12.8

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: cpp-extensionsRelated to torch.utils.cpp_extensionRelated to torch.utils.cpp_extension

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163510 In pytorch/pytorch;

· youkaichao opened on Sep 22, 2025

torch.linalg.ldl_solve aborted heap corruption with wrong input

module: cpuCPU specific problem (e.g., perf, algorithm)CPU specific problem (e.g., perf, algorithm)

module: crashProblem manifests as a hard crash, as opposed to a RuntimeErrorProblem manifests as a hard crash, as opposed to a RuntimeError

module: linear algebraIssues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmulIssues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163450 In pytorch/pytorch;

· qiqicliff opened on Sep 21, 2025

Incorrect bf16 results in Conv

module: bfloat16

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: mkldnnRelated to Intel IDEEP or oneDNN (a.k.a. mkldnn) integrationRelated to Intel IDEEP or oneDNN (a.k.a. mkldnn) integration

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163347 In pytorch/pytorch;

· gvartany-amd opened on Sep 19, 2025

[MPS] scatter_add_ is incorrect on MPS with non-zero offset slices

module: correctness (silent)issue that returns an incorrect result silentlyissue that returns an incorrect result silently

module: mpsRelated to Apple Metal Performance Shaders frameworkRelated to Apple Metal Performance Shaders framework

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163327 In pytorch/pytorch;

· eiennohito opened on Sep 19, 2025

[build] <code>CMake (or ninja)</code> configuration can't index <code>uv</code> environment for incremental compilation

module: buildBuild system issuesBuild system issues

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163311 In pytorch/pytorch;

· shaoyuyoung opened on Sep 19, 2025

[export] Data dependent error on slices in autograd::applySlicing

module: onnxRelated to torch.onnxRelated to torch.onnx

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

oncall: export

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163146 In pytorch/pytorch;

· elin-croft opened on Sep 17, 2025

torch.linalg.eigh uses a large amount of memory in pytorch 2.8.0

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: linear algebraIssues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmulIssues related to specialized linear algebra operations in PyTorch; includes matrix multiply matmul

module: memory usagePyTorch is using more memory than it should, or it is leaking memoryPyTorch is using more memory than it should, or it is leaking memory

needs reproductionEnsure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.Ensure you have actionable steps to reproduce the issue. Someone else needs to confirm the repro.

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#162957 In pytorch/pytorch;

· fjneumann opened on Sep 15, 2025