List view

[CUDA] Upgrade cuDNN to 9.15.1 for CUDA 13 builds

ci-no-tdDo not run TD on this PRDo not run TD on this PR

ciflow/b200

ciflow/binariesTrigger all binary build and upload jobs on the PRTrigger all binary build and upload jobs on the PR

ciflow/h100

ciflow/inductor

ciflow/inductor-periodic

ciflow/inductor-rocmTrigger "inductor" config CI on ROCmTrigger "inductor" config CI on ROCm

ciflow/torchbench

ciflow/trunkTrigger trunk jobs on your pull requestTrigger trunk jobs on your pull request

Merged

Status: Open (in progress).

pytorch/pytorchnumber 169412

#169412 In pytorch/pytorch;

· eqy opened on Dec 2, 2025

Python 3.14 support for PyTorch

module: ciRelated to continuous integrationRelated to continuous integration

module: python frontendFor issues relating to PyTorch's Python frontendFor issues relating to PyTorch's Python frontend

module: python versionIssues related to specific Python versionsIssues related to specific Python versions

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#156856 In pytorch/pytorch;

· atalman opened on Jun 25, 2025

C++ build dependencies

module: cppRelated to C++ APIRelated to C++ API

module: doc infraRelated to pytorch.org/docs, deployment of, and servingRelated to pytorch.org/docs, deployment of, and serving

module: docsRelated to our documentation, both in docs/ and docblocksRelated to our documentation, both in docs/ and docblocks

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.

#163974 In pytorch/pytorch;

· svekars opened on Sep 26, 2025

Setup a notification alert for operator benchmarks regression

module: ciRelated to continuous integrationRelated to continuous integration

module: devxRelated to PyTorch contribution experience (HUD, pytorchbot)Related to PyTorch contribution experience (HUD, pytorchbot)

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.

#164197 In pytorch/pytorch;

· huydhn opened on Sep 30, 2025

[RFC] Cuda support matrix for Release 2.10. Deprecate CUDA 12.6 from CI/CD

oncall: relengIn support of CI and Release EngineeringIn support of CI and Release Engineering

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#165111 In pytorch/pytorch;

· atalman opened on Oct 10, 2025

[RFC] Modifying Getting started page for Experimental Wheel Variant Support

module: docsRelated to our documentation, both in docs/ and docblocksRelated to our documentation, both in docs/ and docblocks

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#166563 In pytorch/pytorch;

· atalman opened on Oct 29, 2025

Significant Memory Regression in <code>F.conv3d</code> with <code>bfloat16</code> Inputs in <code>PyTorch 2.9.0</code>

module: binariesAnything related to official binaries that we release to usersAnything related to official binaries that we release to users

module: convolutionProblems related to convolutions (THNN, THCUNN, CuDNN)Problems related to convolutions (THNN, THCUNN, CuDNN)

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

module: regressionIt used to work, and now it doesn'tIt used to work, and now it doesn't

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.

#166643 In pytorch/pytorch;

· emre-synnada opened on Oct 30, 2025

[RFC] PyTorch - limit supporting poetry same metadata requirements for all packages in download.pytorch.org only

module: binariesAnything related to official binaries that we release to usersAnything related to official binaries that we release to users

topic: binaries

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.

#167658 In pytorch/pytorch;

· atalman opened on Nov 12, 2025

[RFC] Extend MemPool support to SymmMem CUDA backend

module: symm_memIssues and PRs of Symmetric MemoryIssues and PRs of Symmetric Memory

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.

Feature

#167765 In pytorch/pytorch;

· kwen2501 opened on Nov 13, 2025

GradTrackingTensor.tolist() not working on MacOS

module: functorchPertaining to torch.func or pytorch/functorchPertaining to torch.func or pytorch/functorch

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

module: PrivateUse1private useprivate use

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.

#167836 In pytorch/pytorch;

· ysalire opened on Nov 14, 2025

[RFD] Composable Kernel (CK) moving to C++20 — request for feedback/concerns

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#167822 In pytorch/pytorch;

· ThomasNing opened on Nov 14, 2025

On aarch64, <code>pip install torch</code> resulted in the CPU version?

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

module: docsRelated to our documentation, both in docs/ and docblocksRelated to our documentation, both in docs/ and docblocks

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.

#168065 In pytorch/pytorch;

· henrylhtsang opened on Nov 18, 2025

[RFC][torch.compile] Python 3.14 support for <code>torch.compile</code>

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#169875 In pytorch/pytorch;

· williamwen42 opened on Dec 8, 2025

<code>LasrEncoderModelTest::test_sdpa_can_compile_dynamic</code> in <code>transformers</code> fails with torch 2.10.0 nightly but passes with torch 2.9.1

high priority

module: inductor

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.

#169952 In pytorch/pytorch;

· ydshieh opened on Dec 9, 2025

<code>torch.compile</code> fails to capture user-defined kernels with latest Triton

module: inductor

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.

#170049 In pytorch/pytorch;

· jjvraw opened on Dec 10, 2025

[v.2.10.0] Release Tracker

release trackerAdd this label to release tracker issuesAdd this label to release tracker issues

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.

#170119 In pytorch/pytorch;

· atalman opened on Dec 10, 2025

[RFC][Inductor] enablement of combo-kernels - experimental horizontal optimization in torchinductor

module: inductor

oncall: pt2

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#170268 In pytorch/pytorch;

· karthickai opened on Dec 12, 2025

[RFC] DebugMode for tracking dispatched calls and debugging numerical divergence

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#170311 In pytorch/pytorch;

· yushangdi opened on Dec 12, 2025

[Inductor] test_comprehensive_nn_functional_max_pool2d_cuda failed with Triton 3.5

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

upstream tritonUpstream Triton IssueUpstream Triton Issue

Status: Open.

#167339 In pytorch/pytorch;

· desertfire opened on Nov 7, 2025

torch.compile regression 2.9.1: Triton kernel name '_Runner__kernel_name_0' is not defined

actionable

good first issue

module: inductor

module: user tritonrelated to ability to directly torch.compile triton kernelsrelated to ability to directly torch.compile triton kernels

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.

#170398 In pytorch/pytorch;

· dxqb opened on Dec 14, 2025

[Release 2.10] umbrella issue - vLLM CI failures

high priority

module: vllm

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

vllm-compile

Status: Open.

#170433 In pytorch/pytorch;

· atalman opened on Dec 15, 2025

[RFC][Inducor] torch.compile now respects use_deterministic_mode

release-feature-requestThis tag is to mark Feature Tracked for PyTorch OSS ReleasesThis tag is to mark Feature Tracked for PyTorch OSS Releases

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.

#170353 In pytorch/pytorch;

· v0i0 opened on Dec 13, 2025

Enabling CUDA graphs in compile triggers CUDA Invalid Memory Access

high priority

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

module: cuda graphsAbility to capture and then replay streams of CUDA kernelsAbility to capture and then replay streams of CUDA kernels

module: inductor

oncall: pt2

triage review

Status: Open.

#169849 In pytorch/pytorch;

· lw opened on Dec 8, 2025

4x performance regression for 3D convs with AMP on torch 2.9.0

high priority

module: amp (automated mixed precision)autocastautocast

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

module: performanceIssues related to performance, either of kernel code or framework glueIssues related to performance, either of kernel code or framework glue

module: vllm

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.

#166122 In pytorch/pytorch;

· FabianIsensee opened on Oct 23, 2025

[ARM] Unit test regression AArch64 - int4mm kernels require bf16 hw support

module: armRelated to ARM architectures builds of PyTorch. Includes Apple M1Related to ARM architectures builds of PyTorch. Includes Apple M1

module: regressionIt used to work, and now it doesn'tIt used to work, and now it doesn't

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.

#170787 In pytorch/pytorch;

· robert-hardwick opened on Dec 18, 2025