The probabilistic programming language Pyro, initiated by Uber, is now a graduate project of the LF AI & DataFoundation.
The LF AI & Data Foundation, a sub-project of the Linux Foundation, announced that Project Pyro has successfully passed the probationary phase of the Incubator program and is now advancing to the Graduate Level. Pyro is a universal and probabilistic programming language (PPL) originally developed by Uber, which is primarily intended to combine the advantages of deep learning and Bayesian modeling. With the promotion to the graduate project, the Foundation documents both the increasing practical use of Pyro and the growing number of supporters. The programming language is therefore officially considered ready for productive operation.
Pyro is written in Python and tailored to work with the deep learning framework PyTorch. In addition, NumPyro is a JAX-based implementation of the programming language that opens up the same possibilities for flexible and expressive, deeply probabilistic modeling to users of the ML framework JAX. Since Uber handed over responsibility for the open source project to the Linux Foundation in early 2019, applied scientists, statisticians and machine learning experts from universities and research institutions as well as commercial companies – including Siemens and IBM – have been involved in Pyro.
More information on Pyro can be found on the project website where LF AI & Data Foundation and in the GitHub repository. In the current version 1.5.2, Pyro supports Python 3 and is freely available under the Apache 2.0 license.