< FUN.LAB />
// interactive experiments in ML, RL, and creative coding
RL Lunar Lander: Actor-Critic Learns to Land
Land a spacecraft with DQN, A2C, and PPO — all training live in your browser. Watch the actor decide actions while the critic judges states. The foundation of RLHF, visualized. ~4,200 parameters, zero GPUs.
RL Pong: Watch AI Learn to Play in Real Time
Play Pong against Q-Learning, DQN, and REINFORCE agents that train live in your browser. Hit AUTO-TRAIN to watch them practice at 200x speed, or jump in and play against them yourself. 579 parameters, zero GPUs.