Following the simple example of ABC/MSM using Turing.jl (ABC / MSM with Turing.jl), I have updated the https://github.com/mcreel/SimulatedNeuralMoments.jl package to write the likelihood using the Turing.jl PPL form. Sampling is done using AdvancedMH.jl. The code now uses Bijectors.jl to transform the support of the priors, which helps to align the proposal with the likelihood, for more effective sampling. The whole Turing environment works very nicely for this form of ABC/MSM. The Turing version is in the main branch. The released version is still pre-Turing.
Here’s a simple mixture of normals example, to see how it looks:
https://github.com/mcreel/SimulatedNeuralMoments.jl/blob/main/examples/MN/README.md