README.md

metadata

library_name: stable-baselines3
tags:
  - RocketLander-v0
  - deep-reinforcement-learning
  - reinforcement-learning
  - stable-baselines3
model-index:
  - name: TQC
    results:
      - metrics:
          - type: mean_reward
            value: '-0.00 +/- 0.16'
            name: mean_reward
        task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: RocketLander-v0
          type: RocketLander-v0

TQC Agent playing RocketLander-v0

This is a trained model of a TQC agent playing RocketLander-v0 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

Gym env: https://github.com/sdsubhajitdas/Rocket_Lander_Gym

# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo tqc --env RocketLander-v0 -orga araffin -f logs/
python enjoy.py --algo tqc --env RocketLander-v0  -f logs/

Training (with the RL Zoo)

python train.py --algo tqc --env RocketLander-v0 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo tqc --env RocketLander-v0 -f logs/ -orga araffin

Hyperparameters

OrderedDict([('env_wrapper',
              [{'rl_zoo3.wrappers.FrameSkip': {'skip': 4}},
               {'rl_zoo3.wrappers.HistoryWrapper': {'horizon': 2}}]),
             ('n_timesteps', 3000000.0),
             ('policy', 'MlpPolicy'),
             ('normalize', False)])