This repository contains the code used to produce the results of the paper "Differentiable user models".
This code is built on top of the NeuralProcesses.jl library (https://github.com/wesselb/NeuralProcesses.jl). Respectively, we emphasize that the code included in the NeuralProcesses.jl folder in this project does not represent our contribution and is only slightly modified for the purposes of this work.
The ANP model can be trained for the experiment settings introduced in the paper with the following commands:
$ julia --project=Project.toml experiments/ex[n]/experiment[n].jl
where [n]
corresponds to the number of the experiment.
In the first two experiments, the trained models can be straightforwardly evaluated with:
$ julia --project=Project.toml experiments/ex[n]/ex[n]_test.jl
More detailed documentation on the remaining experiments will be added shortly.