Research package for automatic differentiation of programs containing discrete randomness.
142 Stars
Updated Last
1 Year Ago
Started In
September 2022


Build Status arXiv article

StochasticAD is an experimental, research package for automatic differentiation (AD) of stochastic programs. It implements AD algorithms for handling programs that can contain discrete randomness, based on the methodology developed in this NeurIPS 2022 paper. We're still working on docs and code cleanup!


The package can be installed with the Julia package manager:

julia> using Pkg;
julia> Pkg.add("StochasticAD");