PPO LunarLander-v2

PPO agent trained from scratch using PyTorch.

Environment

LunarLander-v2

Training

  • Algorithm: PPO
  • Total timesteps: 50,000
  • Learning rate: 2.5e-4
  • Gamma: 0.99
  • GAE lambda: 0.95
  • Clip coefficient: 0.2

Evaluation

10 evaluation episodes.

  • Mean reward: -157.42
  • Standard deviation: 56.44
  • Certification result: -213.87
Downloads last month
16
Video Preview
loading

Evaluation results