Open-Source Infrastructure
Tools & Code
Active Inference isn't just theory—it's computable. The Institute supports and contributes to an ecosystem of high-performance software enabling probabilistic programming and real-time modeling.
RxInfer.jl
Reactive Message Passing on Factor Graphs
Bayesian Inference at Scale
RxInfer.jl is a powerful Julia package built by the ReactiveBayes ecosystem, specifically designed to automate Bayesian inference using reactive message passing on constraint-based factor graphs.
Within the context of Active Inference, RxInfer allows practitioners to define models via readable Julia macros (powered by GraphPPL.jl), which compile into high-performance message passing networks (powered by ReactiveMP.jl and Rocket.jl).
This hybrid inference engine processes real-time streaming data, enabling seamless simulations of decision-making tasks, multi-agent scenarios, and generative world-models adapting to changing environments.
Institute Repositories
Our GitHub organisation hosts extensive open-source simulation code, documentation infrastructure, and the web components you see here.