These links point to some interesting libraries/projects/repositories for RL algorithms that also include some environments:
- RLLab implements in python very relevant algorithms like Trust Region Policy Optimization (TRPO) and Deep Deterministc Policy Gradient (DDPG) algorithms, Cross-Entropy Method (CEM), Covariance Matrix Adaptation - Evolutionary Strategy (CMA-ES)… and some others.
- Keras-RL implements in python Deep Q-learning (DQN), Double DQN (which removes the bias from max operator in Q-learning), DDPG, Continuous DQN (CDQN or NAF) and CEM.
- BURLAP is Brown University RL and Planning Java library that implements MDP, stochastic games and POMDP and classic and more advance planning algorithms (from A* to Upper Confidence Tree), some standard RL algorithms (from Q-learning to Actor-Critic) and some algorithms for linear value function approximation (Least Squares Policy Iteration, Fitted Value Iteration) and some interesting feature basis functions (tile coding, RBF and Fourier)… among others.
- PyBrain is a general python library for ML that implements classic RL algorithms (Q-Learning and SARSA) and more advance ones (e.g., Neural Fitted Q-iteration and Natural Actor-Critic). It also include a number of black-box policy optimization methods (e.g., CMA-ES, genetic algorithms, etc.).
- Matlab code for book by Richard S. Sutton and Andrew G. Barto.
- ApproxRL: Matlab code for book by Busoniu, Babuska, De Schutter and Ernst.
- PILCO policy search framework using Bayesian optimization.
- Java code for Off-Policy Actor Critic.
- RLPy: linear value function approximation in Python.
Here there are some collections of standard and state-of-the-art environments:
reinforcement learning toolkit - a python implementation of RL by Sutton
http://rlai.cs.ualberta.ca/RLAI/RLtoolkit/RLtoolkit1.0.html
reinforcement learning software and stuff
http://www.cs.ualberta.ca/~sutton/software.html
RL-glue
http://glue.rl-community.org/wiki/RL-Glue_Core
PyBrain - a modular Machine Learning Library for Python
Bayesian reinforcement learning