PyTorch is a free open-source framework that offers an optimized tensor library for deep learning. It offers a number of modules based on the Torch library and can be used for applications such as natural language processing and computer vision. Released in 2016 and developed primarily by Meta (formerly Facebook) AI, deep learning software built on PyTorch include Tesla Autopilot and Uber’s Pyro.
PyTorch offers two high-level features:
- Tensor computing that can take advantage of acceleration via GPUs.
- Deep neural networks.
Instructions for installing on various systems can be found on pytorch.org.
NumPymodule is a prerequisite for a PyTorch install.