TellWorldAI World Archive

Home›News›Release

ReleaseDeep learning

PyTorch is publicly released

PyTorch was publicly released in January 2017 as a beta, with first-class Python integration and an imperative style. Built by people from the Lua Torch community with Facebook's backing, it later became a widely used deep-learning framework.

Laptop and circuit board on a workbench

Background

Deep-learning researchers needed software that could build and train large neural networks. PyTorch's development began in 2016 as a collective effort by people from the Lua Torch community. A large share of the people and funding came from Facebook, and contributors came from NVIDIA, Twitter and other organisations.[2]

What happened

PyTorch was publicly released as a beta in January 2017. Its release notes give 18 January as the public release date.[1] Its original authors were Adam Paszke, Sam Gross, Soumith Chintala and Gregory Chanan.[3]

Key details

PyTorch's strengths were first-class Python integration and an imperative programming style. Versions 0.2 to 0.4 followed over the next year.[4]

In 2018 Facebook merged Caffe2 into PyTorch. PyTorch 1.0 shipped on 7 December 2018, with the torch.jit compiler and a C++ frontend.[4][5]

OpenAI standardised on PyTorch in January 2020.[3]

By September 2022 more than 150,000 GitHub projects were built on PyTorch.[6]

A deep-learning framework provides the tools for defining neural networks, computing gradients and running training on hardware such as GPUs. An imperative style means each line runs when it is reached, which makes models easier to inspect and debug.[4]

PyTorch's roots lie in the Lua Torch community, where its first developers came from.[2] Python was widely used among machine-learning researchers, so good integration with it mattered to them. The imperative style also suited the way researchers wrote and tested experiments step by step.[4]

The public beta made the framework available for anyone to download and use. Its development was carried out by a group of contributors from several organisations, not by a single company alone.[3][2]

Impact

PyTorch became one of the main frameworks through which researchers and companies build deep-learning systems. The release also marks the start of a project that Meta later handed to an independent foundation, as described in a separate entry in this archive.[6][2]

Sources

  1. 1
  2. 2
  3. 3
    PyTorchWikipedia
  4. 4
  5. 5
  6. 6

Cite this page: PyTorch is publicly released. TellWorld, https://www.tellworld.net/news/pytorch-2017. Updated 2026-11-08

Related news

View all