tanu20/ppo-LunarLander-v2-unit8

PPO LunarLander-v2

PPO agent implemented from scratch using PyTorch for Unit 8 Part I of the Hugging Face Deep Reinforcement Learning Course.

Evaluation

Mean reward: -132.37 +/- 43.43

Result used by the certification checker:

-175.80

Evaluation episodes: 10

Algorithm

• Proximal Policy Optimization (PPO)
• Actor-Critic
• Generalized Advantage Estimation (GAE)
• PPO clipped objective
• PyTorch

Environment

The original Unit 8 course uses LunarLander-v2.

The current Gymnasium release has deprecated v2 and provides LunarLander-v3, so evaluation was performed with the current compatible environment while preserving the course's required LunarLander-v2 certification metadata.

Files

• model.pt
• results.json
• replay.mp4

Course

Hugging Face Deep Reinforcement Learning Course Unit 8 Part I — PPO