pyro-ppl/pyro

Health v2 · maintenance only

Adoption not part of the score

9037 stars · 1015 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Pyro is a deep universal probabilistic programming library built on Python and PyTorch, supporting Bayesian modeling with variational inference, MCMC, and other scalable inference algorithms. It can represent any computable probability distribution and integrates with deep learning workflows.

Use cases

• fit bayesian models with variational inference in python
• run MCMC sampling on a probabilistic model with PyTorch
• build a variational autoencoder with uncertainty estimates
• do bayesian hierarchical regression
• model time series with probabilistic forecasting
• estimate uncertainty in deep neural network predictions

When to choose

• you want a flexible, universal PPL backed by PyTorch and GPU acceleration
• you need both automated and customizable inference (SVI, HMC/NUTS, SMC)
• you want to combine deep learning with Bayesian modeling
• you need to scale probabilistic models to large datasets

When to avoid

• you want maximum HMC/NUTS speed and JIT compilation - consider NumPyro with JAX instead
• you need simple classical statistics without probabilistic programming
• your project is not Python/PyTorch based

Facets

library · maturity stable

machine-learning deep-learning math data-science machine-learning deep-learning data-science python cross-platform probabilistic-programming bayesian-inference variational-inference mcmc pytorch generative-models algorithms gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="pyro-ppl/pyro")