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.

distributed, likelihood-free inference

A system for scientific simulation-based inference at scale.

Likelihood-free AMortized Posterior Estimation with PyTorch

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

Roundtrip: density estimation with deep generative neural networks

Lectures on Bayesian statistics and information theory

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

Julia package for neural estimation

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

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

Evaluating model calibration methods for sensitivity analysis, uncertainty analysis, optimisation, and Bayesian inference

Simulator Expansion for Likelihood-Free Inference (SELFI): a python implementation

Likelihood-Free Inference for Julia.

Code and manuscript for the paper "INFERNO: Inference-Aware Neural Optimisation". Automated mirror from CERN GitLab.

Correlation functions versus field-level inference in cosmology: example with log-normal fields

PyTorch implementation of inference aware neural optimisation (de Castro and Dorigo, 2018 https://www.sciencedirect.com/science/article/pii/S0010465519301948)

A simulation model for the digital reconstruction of 3D root system architectures. Integrated with a simulation-based inference generative deep learning model.

Simulator of the Lotka-Volterra prey-predator system with demographic and observational noise and biases

Code for "Neural Likelihood Surfaces for Spatial Processes with Computationally Intensive or Intractable Likelihoods" (arxiv:2305.04634)