Jaxplorer is a Jax reinforcement learning (RL) framework for exploring new ideas.

Implementation of Continuous Control RL Algorithms

The implementation for the paper Continuous Deep Q-Learning in Optimal Control Problems: Normalized Advantage Functions Analysis // NeurIPS 2022

Reinforcement learning, robotics simulation with agent using NAF+HER learning method. Agent's task is to fetch puck to desired position using dynamic contact.

lightweight lxml wrapper for NAF

some runscripts for usage on the NAF and other useful scripts