🚀 PPO LunarLander-v3
A Proximal Policy Optimization (PPO) agent trained with Stable-Baselines3 to solve the LunarLander-v3 environment from Gymnasium.
The agent was trained for 3.5 million timesteps and evaluated over 100 episodes using deterministic actions.
🎥 Watch the agent in action
The trained agent's replay is available in the Video Preview panel on the right.
📊 Evaluation Results
The trained agent was evaluated for 100 episodes.
| Metric | Result |
|---|---|
| 🏆 Mean Reward | 258.01 |
| 📊 Standard Deviation | 20.59 |
| ⬆️ Best Reward | 291.68 |
| ⬇️ Worst Reward | 216.40 |
| 🎯 Evaluation Episodes | 100 |
| 🔥 Training Timesteps | 3,500,000 |
| 📈 Lower Bound | 237.42 |
Leaderboard Score
The Deep Reinforcement Learning Course leaderboard uses:
mean_reward - std_reward
- Downloads last month
- 43
Evaluation results
- mean_reward on LunarLander-v3self-reported258.010
- std_reward on LunarLander-v3self-reported20.590