README.md

metadata

tags:
  - LunarLander-v2
  - ppo
  - deep-reinforcement-learning
  - reinforcement-learning
  - custom-implementation
  - deep-rl-course
model-index:
  - name: PPO
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: LunarLander-v2
          type: LunarLander-v2
        metrics:
          - type: mean_reward
            value: 104.83 +/- 18.01
            name: mean_reward
            verified: false

PPO Agent Playing LunarLander-v2

This is a trained model of a PPO agent playing LunarLander-v2.

Hyperparameters

See the GitHub for full info and the journey on creating this on the surface not particularly exciting model: https://github.com/MattStammers/PPO_Lander_Implementation

It took me 8 attempts to get the score to nearly reach 0 using a cleanRL implementation and WandB metric tracking and then this version was trained after 10 attempts converging at about 3 million training steps