|
--- |
|
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: -43.48 +/- 111.28 |
|
name: mean_reward |
|
verified: false |
|
--- |
|
|
|
# PPO Agent Playing LunarLander-v2 |
|
|
|
This is a trained model of a PPO agent playing LunarLander-v2. |
|
|
|
# Hyperparameters |
|
```python |
|
{'exp_name': 'ppo' |
|
'seed': 3 |
|
'torch_deterministic': True |
|
'cuda': True |
|
'track': False |
|
'wandb_project_name': 'cleanRL' |
|
'wandb_entity': None |
|
'capture_video': False |
|
'env_id': 'LunarLander-v2' |
|
'total_timesteps': 100000 |
|
'learning_rate': 0.00025 |
|
'num_envs': 4 |
|
'num_steps': 128 |
|
'anneal_lr': True |
|
'gae': True |
|
'gamma': 0.93 |
|
'gae_lambda': 0.9 |
|
'num_minibatches': 4 |
|
'update_epochs': 8 |
|
'norm_adv': True |
|
'clip_coef': 0.15 |
|
'clip_vloss': True |
|
'ent_coef': 0.01 |
|
'vf_coef': 0.5 |
|
'max_grad_norm': 0.5 |
|
'target_kl': None |
|
'repo_id': 'MoeenTB/ppo-LunarLander-v2_scratch_attempt12' |
|
'batch_size': 512 |
|
'minibatch_size': 128} |
|
``` |
|
|
