MDPs and tabular methods: finite MDPs with explicit transition tables; policy evaluation, policy iteration and value iteration; Monte Carlo, TD(0), SARSA and Q-learning.
Environments: GridWorld, FrozenLake, CliffWalking, Taxi, CartPole, MountainCar, Acrobot and Pendulum, checked step by step against Gymnasium 0.29.1; synchronous vector environments, frame stacking, observation and reward normalisation.
Deep RL on Owl's neural networks: linear function approximation, REINFORCE (with an optional baseline), DQN, A2C and PPO, with categorical and Gaussian policies. DQN and PPO match CleanRL's published results on its benchmark environments.
RL for language models (owl-rllib.llm): a toy autoregressive model on two-digit addition, with REINFORCE, RLOO, token-level PPO, GRPO and the GRPO variants Dr. GRPO, DAPO, GSPO and CISPO.
Python bridge (owl-rllib.bridges, optional): Gymnasium environments and VizDoom scenarios run in a Python child process.