Posteriordb v. 1.1.0 released

General
1

We are happy to announce the release of posteriordb 1.1.

posteriordb is a database of Bayesian posterior inference problems, containing models, datasets, posterior definitions, and reference posterior draws. It is intended for testing, benchmarking, and developing Bayesian inference algorithms and probabilistic programming tools.

This release expands PosteriorDB with broader PyMC support, additional reference posterior draws, new summary statistics, and several model/specification fixes.

Highlights:

  • Added 73 PyMC model implementations under posterior_database/models/pymc.
  • Added the new posterior/model variant M0_data-M0_model_discrete.
  • Added 11 new reference posterior draw objects, increasing the total from 46 to 57.
  • Added corresponding mean_value and mean_squared_value summary statistics.
  • Updated model metadata with PyMC implementation information and version metadata.
  • Updated citation metadata to the AISTATS 2025 PosteriorDB paper.
  • Added Croissant metadata and updated licensing information.
  • Fixed or aligned several model specifications, including Covid, dogs_hierarchical, selected diamonds, earnings, kidiq, mesquite, and IRT models.

The release was checked with the PosteriorDB content validation, and all 120 Stan model files parse successfully with rstan::stanc().

Release:

Repository:

Paper:

Thanks to everyone who contributed models, reference draws, fixes, and reviews!

5 Likes