(CleanRL) C51 Agent Playing PongNoFrameskip-v4
This is a trained model of a C51 agent playing PongNoFrameskip-v4. 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[c51_atari]"
python -m cleanrl_utils.enjoy --exp-name c51_atari --env-id PongNoFrameskip-v4
Please refer to the documentation for more detail.
Command to reproduce the training
curl -OL https://huggingface.co/kinalmehta/PongNoFrameskip-v4-c51_atari-seed1/raw/main/c51_atari.py
curl -OL https://huggingface.co/kinalmehta/PongNoFrameskip-v4-c51_atari-seed1/raw/main/pyproject.toml
curl -OL https://huggingface.co/kinalmehta/PongNoFrameskip-v4-c51_atari-seed1/raw/main/poetry.lock
poetry install --all-extras
python c51_atari.py --save-model --upload-model --hf-entity kinalmehta --env-id PongNoFrameskip-v4
Hyperparameters
{'batch_size': 32,
'buffer_size': 1000000,
'capture_video': False,
'cuda': True,
'end_e': 0.01,
'env_id': 'PongNoFrameskip-v4',
'exp_name': 'c51_atari',
'exploration_fraction': 0.1,
'gamma': 0.99,
'hf_entity': 'kinalmehta',
'learning_rate': 0.00025,
'learning_starts': 80000,
'n_atoms': 51,
'save_model': True,
'seed': 1,
'start_e': 1,
'target_network_frequency': 10000,
'torch_deterministic': True,
'total_timesteps': 10000000,
'track': False,
'train_frequency': 4,
'upload_model': True,
'v_max': 10,
'v_min': -10,
'wandb_entity': None,
'wandb_project_name': 'cleanRL'}
Evaluation results
- mean_reward on PongNoFrameskip-v4self-reported18.00 +/- 1.00