RyanAA/pt1-ppo-LunarLander-v2

PPO Agent Playing LunarLander-v2

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

I've included the course ppo.py file as well.

Hyperparameters

--total-timesteps=500000
--learning-rate=0.0003
--num-envs=8
--num-steps=1024
--num-minibatches=32
--update-epochs=10
--gamma=0.999
--gae-lambda=0.98
--clip-coef=0.2
--ent-coef=0.01
--vf-coef=0.5
--max-grad-norm=0.5