metadata
tags:
- Breakout-v5
- deep-reinforcement-learning
- reinforcement-learning
- custom-implementation
library_name: cleanrl
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Breakout-v5
type: Breakout-v5
metrics:
- type: mean_reward
value: 807.50 +/- 137.36
name: mean_reward
verified: false
(CleanRL) PPO Agent Playing Breakout-v5
This is a trained model of a PPO agent playing Breakout-v5. The model was trained by using CleanRL and the most up-to-date training code can be found here.
Get Started
To use this model, please install the cleanrl
package with the following command:
pip install "cleanrl[jax,envpool,atari]"
python -m cleanrl_utils.enjoy --exp-name sebulba_ppo_envpool_impala_atari_wrapper --env-id Breakout-v5
Please refer to the documentation for more detail.
Command to reproduce the training
curl -OL https://huggingface.co/cleanrl/Breakout-v5-sebulba_ppo_envpool_impala_atari_wrapper-seed2/raw/main/sebulba_ppo_envpool_impala_atari_wrapper.py
curl -OL https://huggingface.co/cleanrl/Breakout-v5-sebulba_ppo_envpool_impala_atari_wrapper-seed2/raw/main/pyproject.toml
curl -OL https://huggingface.co/cleanrl/Breakout-v5-sebulba_ppo_envpool_impala_atari_wrapper-seed2/raw/main/poetry.lock
poetry install --all-extras
python sebulba_ppo_envpool_impala_atari_wrapper.py --actor-device-ids 0 --learner-device-ids 1 2 3 4 5 6 --track --save-model --upload-model --hf-entity cleanrl --env-id Breakout-v5 --seed 2
Hyperparameters
{'actor_device_ids': [0],
'anneal_lr': True,
'async_batch_size': 20,
'async_update': 3,
'batch_size': 7680,
'capture_video': False,
'clip_coef': 0.1,
'cuda': True,
'ent_coef': 0.01,
'env_id': 'Breakout-v5',
'exp_name': 'sebulba_ppo_envpool_impala_atari_wrapper',
'gae_lambda': 0.95,
'gamma': 0.99,
'hf_entity': 'cleanrl',
'learner_device_ids': [1, 2, 3, 4, 5, 6],
'learning_rate': 0.00025,
'max_grad_norm': 0.5,
'minibatch_size': 1920,
'norm_adv': True,
'num_actor_threads': 1,
'num_envs': 60,
'num_minibatches': 4,
'num_steps': 128,
'num_updates': 6510,
'profile': False,
'save_model': True,
'seed': 2,
'target_kl': None,
'test_actor_learner_throughput': False,
'torch_deterministic': True,
'total_timesteps': 50000000,
'track': True,
'update_epochs': 4,
'upload_model': True,
'vf_coef': 0.5,
'wandb_entity': None,
'wandb_project_name': 'cleanRL'}