Rate this Page

PyTorch Governance | Maintainers#

Responsibilities#

  • Triage and fix high priority issues assigned to the module or library

  • Triage, review, and land high priority pull requests assigned to the module or library

  • Answer module or library questions on discuss.pytorch.org and dev-discuss.pytorch.org

  • Maintain public user and development documentation

  • Run meetings and share minutes plus roadmap on a half or quarterly basis

Lead Core Maintainer (BDFL)#

Core Maintainers#

Module-level maintainers#

NN APIs (torch.nn)#

Optimizers (torch.optim)#

Autograd (torch.autograd)#

TorchDynamo#

TorchInductor#

Cudagraph Tree#

PT2 Dispatcher#

PT2 Export (torch.export)#

AOT Inductor (AOTI) & AOTI Runtime#

Compilers (JIT / TorchScript / Package / Deploy)#

Distributions & RNG#

Distributed#

Multiprocessing#

Linear Algebra (torch.linalg)#

Sparse (torch.sparse)#

NestedTensor (torch.nested)#

MaskedTensor (torch.masked)#

Fast Fourier Transform (torch.fft)#

MKLDNN#

CUDA#

AMD/ROCm/HIP#

Build + CI#

Performance Tools#

C++ API#

C10 utils and operator dispatch#

ONNX exporter#

LiteInterpreter#

Quantization (torch/ao)#

Windows#

Apple M1/MPS/Metal#

PowerPC#

x86 CPU#

AArch64 CPU#

Docs / Tutorials#

Library-level maintainers#

XLA#

TorchServe#

TorchVision#

TorchText#

TorchAudio#

TorchRec#

TorchX#

TorchData#

TorchArrow#

ExecuTorch (Edge, Mobile)#

TorchTune#

TorchChat#

TorchCodec#