Examples

Interactive notebooks demonstrating each algorithm in the lit-sb3 library. All examples use the SB3-compatible API backed by PyTorch Lightning.

---

Introduction

• Introduction notebook

---

On-Policy Algorithms

These algorithms collect a rollout, compute updates, then discard the data.

Notebook	Algorithm	Environment
A2C on CartPole-v1	Advantage Actor-Critic	CartPole-v1 (discrete)
PPO on CartPole-v1	Proximal Policy Optimization	CartPole-v1 (discrete)

---

Off-Policy Algorithms

These algorithms store transitions in a replay buffer and can reuse past experience.

Notebook	Algorithm	Environment
DQN on CartPole-v1	Deep Q-Network	CartPole-v1 (discrete)
SAC on LunarLanderContinuous-v2	Soft Actor-Critic	LunarLanderContinuous-v2 (continuous)
DDPG on LunarLanderContinuous-v2	Deep Deterministic Policy Gradient	LunarLanderContinuous-v2 (continuous)
TD3 on LunarLanderContinuous-v2	Twin Delayed DDPG	LunarLanderContinuous-v2 (continuous)