The State of the Art for Probabilistic Programming

on YouTube ↗47m

" summary

Survey of concrete advances and demos: PyMC moves from monolith to layered stack, compiling to PyTensor and BlackJAX or Rust samplers; examples show JAX and a Rust NUTS reducing run time dramatically (one demo reports 70× speedup). Browser proofs run via PyScript and pyodide. Features implemented or experimental include auto-marginalization, Bayesian causal workflows (CausalPy) and interoperable backends; most demos work but some cases remain limited.

AI-written summary · about

" chapters (25)

1. 0:00Introduction by the moderator
2. 0:38Thomas introduces himself
3. 3:20Probabilistic programming GitHub star history
4. 5:17PyTensor Announcement
5. 7:18The Great Uncoupling of packages
6. 12:44BlackJAX: Inference Algorithms in JAX
7. 15:40Nutpie sampler: NUTS written in Rust
8. 15:51PyScript and pyodide
9. 19:01Demonstration of how PyScript and pyodide work
10. 20:26Auto-marginalization
11. 22:07Bayesian Causal Inference
12. 23:18CausalPy
13. 24:15Bayes in business trends
14. 26:30PyMCon Web series: https://pymcon.com/
15. 27:06Intuitive Bayes course
16. 29:04Q & A Are you working with banks or insurers on capital modeling?
17. 30:33Q & A Is there an overlap between Aesara and JAX?
18. 32:50Q & A How general is the auto-marginalization, can it work for arbitrary distributions?
19. 34:06Q & A Does PyMC have any functionality for time series models yet?
20. 35:36Q & A Do you encourage and work with external bodies to contribute to the PyMC universe?
21. 38:13Q & A It feels like the library is moving fast, is the documentation and learning resources or examples keeping the same fast pace?
22. 41:00Q & A If one wanted to contribute to PyMC, is it beginner friendly?
23. 44:37Q & A Will the video of the talk be shared?
24. 46:44How to get in touch with PyMC Labs
25. 47:09Thank you!

" related talks

• 43mProbabilistic Programming in the Real WorldZach Anglin · PyData DC 2018
• 01h 27mProbabilistic Python: An Introduction to Bayesian Modeling with PyMCChris Fonnesbeck · PyData London 2022
• 01h 03mAn Overview of Probabilistic ProgrammingVikash K. Mansinghka · Strange Loop 2015
• 43mThe Power of Probabilistic ProgrammingBen Vigoda · O'Reilly Artificial Intelligence Conference 2017 - New York, New York
• 01h 07mProbabilistic Programming and Readable ModelsFritz Obermeyer · PyData Yerevan 2022
• 01h 26mProbabilistic Programming and Bayesian Computing with PyMCFonnesbeck, Wiecki · PyData London 2024
• 01h 31mProbabilistic Programming and Bayesian Inference in PythonLara Kattan · PyOhio 2019
• 40mProbabilistic Programming for Augmented IntelligenceVikash K. Mansinghka · Uncertainty in Computation 2016