Learning Optimal Deterministic Policies with Stochastic Policy Gradients

This is the code used for the paper Learning Optimal Deterministic Policies with Stochastic Policy Gradients, published at ICML 2024.

In particular, it contains the implementation of the GPOMDP and PGPE algorithms, returning deterministic policies and running over the principal MuJoCo environments and on some other hand-crafted ones, as ContinuousCartPole and LQR.