Research-oriented reinforcement learning library offering high-quality single-file implementations of algorithms like PPO, DQN, and SAC, with benchmarking and large-scale experiment support.