Quentin Gallouédec
Initial commit
22b86a8
metadata
library_name: stable-baselines3
tags:
  - HalfCheetah-v3
  - deep-reinforcement-learning
  - reinforcement-learning
  - stable-baselines3
model-index:
  - name: ARS
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: HalfCheetah-v3
          type: HalfCheetah-v3
        metrics:
          - type: mean_reward
            value: 6015.46 +/- 163.71
            name: mean_reward
            verified: false

ARS Agent playing HalfCheetah-v3

This is a trained model of a ARS agent playing HalfCheetah-v3 using the stable-baselines3 library and the RL Zoo.

The RL Zoo is a training framework for Stable Baselines3 reinforcement learning agents, with hyperparameter optimization and pre-trained agents included.

Usage (with SB3 RL Zoo)

RL Zoo: https://github.com/DLR-RM/rl-baselines3-zoo
SB3: https://github.com/DLR-RM/stable-baselines3
SB3 Contrib: https://github.com/Stable-Baselines-Team/stable-baselines3-contrib

Install the RL Zoo (with SB3 and SB3-Contrib):

pip install rl_zoo3
# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo ars --env HalfCheetah-v3 -orga qgallouedec -f logs/
python -m rl_zoo3.enjoy --algo ars --env HalfCheetah-v3  -f logs/

If you installed the RL Zoo3 via pip (pip install rl_zoo3), from anywhere you can do:

python -m rl_zoo3.load_from_hub --algo ars --env HalfCheetah-v3 -orga qgallouedec -f logs/
python -m rl_zoo3.enjoy --algo ars --env HalfCheetah-v3  -f logs/

Training (with the RL Zoo)

python -m rl_zoo3.train --algo ars --env HalfCheetah-v3 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo ars --env HalfCheetah-v3 -f logs/ -orga qgallouedec

Hyperparameters

OrderedDict([('alive_bonus_offset', 0),
             ('delta_std', 0.03),
             ('learning_rate', 0.02),
             ('n_delta', 32),
             ('n_envs', 1),
             ('n_timesteps', 12500000.0),
             ('n_top', 4),
             ('normalize', 'dict(norm_obs=True, norm_reward=False)'),
             ('policy', 'LinearPolicy'),
             ('normalize_kwargs', {'norm_obs': True, 'norm_reward': False})])