PPO Agent Playing LunarLander-v2
This is a trained model of a PPO agent playing LunarLander-v2.
Hyperparameters
{'exp_name': 'ppo'
'gym_id': 'LunarLander-v2'
'learning_rate': 0.00025
'seed': 1
'total_timesteps': 1000000
'torch_deterministic': True
'cuda': True
'capture_video': False
'num_envs': 4
'num_steps': 128
'batch_size': 512
'anneal_lr': True
'gae': True
'gamma': 0.99
'gae_lambda': 0.95
'num_minibatches': 4
'update_epochs': 4
'normalize_advantages': True
'clip_coefficient': 0.2
'clip_value_loss': True
'entropy_coefficient': 0.01
'vf_coefficient': 0.5
'max_gradient_norm': 0.5
'repo_id': 'felixb85/ppo-LunarLander-v2'
'minibatch_size': 128
'env_id': 'LunarLander-v2'}
Evaluation results
- mean_reward on LunarLander-v2self-reported57.03 +/- 121.77