README.md

metadata

tags:
  - HalfCheetah-v4
  - deep-reinforcement-learning
  - reinforcement-learning
  - custom-implementation
library_name: cleanrl
model-index:
  - name: PPO
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: HalfCheetah-v4
          type: HalfCheetah-v4
        metrics:
          - type: mean_reward
            value: 1731.91 +/- 19.92
            name: mean_reward
            verified: false

(CleanRL) PPO Agent Playing HalfCheetah-v4

This is a trained model of a PPO agent playing HalfCheetah-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[ppo_fix_continuous_action]"
python -m cleanrl_utils.enjoy --exp-name ppo_fix_continuous_action --env-id HalfCheetah-v4

Please refer to the documentation for more detail.

Command to reproduce the training

curl -OL https://huggingface.co/sdpkjc/HalfCheetah-v4-ppo_fix_continuous_action-seed3/raw/main/ppo_fix_continuous_action.py
curl -OL https://huggingface.co/sdpkjc/HalfCheetah-v4-ppo_fix_continuous_action-seed3/raw/main/pyproject.toml
curl -OL https://huggingface.co/sdpkjc/HalfCheetah-v4-ppo_fix_continuous_action-seed3/raw/main/poetry.lock
poetry install --all-extras
python ppo_fix_continuous_action.py --save-model --upload-model --hf-entity sdpkjc --env-id HalfCheetah-v4 --seed 3 --track

Hyperparameters

{'anneal_lr': True,
 'batch_size': 2048,
 'capture_video': False,
 'clip_coef': 0.2,
 'clip_vloss': True,
 'cuda': True,
 'ent_coef': 0.0,
 'env_id': 'HalfCheetah-v4',
 'exp_name': 'ppo_fix_continuous_action',
 'gae_lambda': 0.95,
 'gamma': 0.99,
 'hf_entity': 'sdpkjc',
 'learning_rate': 0.0003,
 'max_grad_norm': 0.5,
 'minibatch_size': 64,
 'norm_adv': True,
 'num_envs': 1,
 'num_minibatches': 32,
 'num_steps': 2048,
 'save_model': True,
 'seed': 3,
 'target_kl': None,
 'torch_deterministic': True,
 'total_timesteps': 1000000,
 'track': True,
 'update_epochs': 10,
 'upload_model': True,
 'vf_coef': 0.5,
 'wandb_entity': None,
 'wandb_project_name': 'cleanRL'}