Post
jaxdp
A Dynamic Programming package for finite MDPs implemented in JAX
Jaxdp is a Python package that provides simple functional implementation of dynamic programming (DP) algorithms for discrete state-action Markov decision processes (MDP) within the JAX ecosystem. Using the JAX transformations, you can accelerate (even using GPUs) DP algorithms by running multiple MDPs, initial values etc. in a vectorized form.
Post
mjnax
MuJoCo-Based Control Environments Compatible with Gymnax.
mjnax is a library of control and robotics environments designed to be compatible with the gymnax API. Built on MuJoCo-MJX, mjnax enables environments to run efficiently on GPUs and supports automatic vectorization.
Post
modular baselines
Deep reinforcement learning package
Modular-Baselines is a Reinforcement Learning (RL) library, based on Stable-Baselines3, with the objective of improving flexibility and providing necessary components in RL Research. Components are framework agnostic in the sense that they do not rely on a specific framework. That said, Modular baselines includes Pytorch implementation of some of the agents.
