A category of posts focused on production usage of PyTorch. Mobile deployment is out of scope for this category (for now… )

A category of posts relating to the autograd engine itself.

Topics related to either pytorch/vision or vision research related topics

A category for torch.compile and PyTorch 2.0 related compiler issues.
This includes: issues around TorchDynamo ( torch._dynamo ), TorchInductor (torch._inductor) and AOTAutograd

Topics related to DataLoader, Dataset, torch.utils.data, pytorch/data, and TorchArrow.

This category is focused on PyTorch on Windows related issues.

Topics related to the C++ Frontend, C++ API or C++ Extensions

This category is for questions, discussion and issues related to PyTorch’s quantization feature.

Topics related to Natural Language Processing

A section to discuss RL implementations, research, problems

A category of posts relating to ExecuTorch.

This category is for the Glow neural network accelerator compiler: https://github.com/pytorch/glow

A category for TorchScript and the PyTorch JIT compiler

Please redirect your questions to https://github.com/pytorch/opacus; we are not able to provide any guarantee on response time to Opacus questions on the PyTorch forums.

This category is to discuss xla/TPU related issues.

PyTorch Live is no longer supported. Please look into ExecuTorch as the new Mobile runtime for PyTorch.

This category is for any question related to MPS support on Apple hardware (both M1 and x86 with AMD machines).

this category is focused on python deployment of PyTorch models and specifically the torch::deploy and torch.package APIs. More can be found at pytorch.org in the docs…

This category is dedicated to the now deprecated “PyTorch Mobile” project. Please look into ExecuTorch as the new Mobile runtime for PyTorch.

Tell the community how you’re using PyTorch!

Discussion about this site, its organization, how it works, and how we can improve it.

TorchX is an SDK for quickly building and deploying ML applications from R&D to production. It offers various builtin components that encode MLOps best practices and make advanced features like distributed training and hyperparameter optimization accessible to all. Users can get started with TorchX with no added setup cost since it supports popular ML schedulers and pipeline orchestrators that are already widely adopted and deployed in production.

The FAQ category contains commonly-asked questions and their answers. Please refer to this section before you post your query.

Use this category to discuss ideas about the PyTorch Global and local Hackathons.

torchchat - Running LLMs locally