sbi is a Python package for simulation-based inference, designed to meet the needs of both researchers and practitioners. Whether you need fine-grained control or an easy-to-use interface, sbi has you covered.

A system for scientific simulation-based inference at scale.

Likelihood-free AMortized Posterior Estimation with PyTorch

R package for statistical inference using partially observed Markov processes

Community-sourced list of papers and resources on neural simulation-based inference.

Density estimation likelihood-free inference. No longer actively developed see https://github.com/mackelab/sbi instead

A Python toolkit for (simulation-based) inference and the mechanization of science.

Automatically convert Julia methods to Gen functions.

Simulation-based inference in JAX

Simulation-based (likelihood-free) inference customized for astronomical applications

Fast Bayesian optimization, quadrature, inference over arbitrary domain with GPU parallel acceleration

SBI Workshop jointly by Helmholtz AI + ML ⇌ Science Colaboratory

(NeurIPS 2022) Fast Bayesian Inference with Batch Bayesian Quadrature via Kernel Recombination

Probing the nature of dark matter by inferring the dark matter particle mass with machine learning and stellar streams.

Julia package for neural estimation

Fast, lightweight and parallelised simulation-based inference in JAX.

Normalizing flow models allowing for a conditioning context, implemented using Jax, Flax, and Distrax.

Conduct simulation-based inference on strong gravitational lensing systems.

Code for the paper "Towards Reliable Simulation-Based Inference with Balanced Neural Ratio Estimation".