ONNX Probabilistic Programming Working Group – Announcement & Call for Contributors

Publicity Events
1

Hi everyone,

I wanted to share that the ONNX Probabilistic Programming Working Group has officially been formed. The goal of this working group is to bring probabilistic modeling and Bayesian inference into the ONNX ecosystem as first-class capabilities.

The intent is to define a standardized operator domain and runtime semantics that allow probabilistic models to be exported, executed, and optimized across frameworks and hardware, similar to what ONNX has already enabled for neural networks.

A major focus of this effort is ensuring that probabilistic semantics are preserved across ecosystems, and the Stan ecosystem is an important reference point for this work given its strong foundation in Bayesian modeling, inference algorithms (HMC/NUTS), and numerical stability.

Some of the areas the working group will be exploring include:

  • Probabilistic operator domains (distributions, log-probabilities, factors)
  • Bijectors and parameter constraints
  • Reproducible stateless RNG semantics
  • Special mathematical functions used in probabilistic computation
  • Inference algorithms such as Laplace, Pathfinder, INLA, HMC, NUTS, and SMC
  • Exporter pathways for probabilistic programming frameworks including Stan, PyMC, Pyro, NumPyro, TensorFlow Probability, JAX-based systems, BayesFlow, and future Julia/Turing support

The long-term goal is to enable probabilistic models to be portable across frameworks and hardware backends using ONNX as an intermediate representation.

If you’re interested in participating or providing feedback from the Stan community perspective, we would love to hear from you.

You can reach out to:

You are also welcome to attend the working group meetings:

Fridays @ 12 PM EST, every two weeks

Working group repository:

2 Likes