PyTorch was developed by Facebook’s AI research group and released as an open-source machine-learning framework.
Facebook Artificial Intelligence Research, known as FAIR, introduced PyTorch in 2016. The framework was built around Python and the Torch machine-learning library, with strong support for tensor computation and neural-network development.
PyTorch became widely used in research because its dynamic computational-graph approach made experimentation and debugging relatively flexible. It was also used in computer vision, natural-language processing, and other artificial-intelligence applications.
PyTorch is often confused with TensorFlow, another major machine-learning framework. TensorFlow was developed by the Google Brain team. PyTorch’s corporate origin is Facebook, although the project later moved under the Linux Foundation in 2022.