README.md

metadata

library_name: stable-baselines3
tags:
  - PandaPush-v1
  - deep-reinforcement-learning
  - reinforcement-learning
  - stable-baselines3
  - arxiv.org/abs/2106.13687
model-index:
  - name: TQC
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: PandaPush-v1
          type: PandaPush-v1
        metrics:
          - type: mean_reward
            value: '-14.60 +/- 16.30'
            name: mean_reward
            verified: false

TQC Agent playing PandaPush-v1

This is a trained model of a TQC agent playing PandaPush-v1 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 tqc --env PandaPush-v1 -orga qgallouedec -f logs/
python -m rl_zoo3.enjoy --algo tqc --env PandaPush-v1  -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 tqc --env PandaPush-v1 -orga qgallouedec -f logs/
python -m rl_zoo3.enjoy --algo tqc --env PandaPush-v1  -f logs/

Training (with the RL Zoo)

python -m rl_zoo3.train --algo tqc --env PandaPush-v1 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo tqc --env PandaPush-v1 -f logs/ -orga qgallouedec

Hyperparameters

OrderedDict([('batch_size', 2048),
             ('buffer_size', 1000000),
             ('env_wrapper', 'sb3_contrib.common.wrappers.TimeFeatureWrapper'),
             ('gamma', 0.95),
             ('learning_rate', 0.001),
             ('n_timesteps', 1000000.0),
             ('policy', 'MultiInputPolicy'),
             ('policy_kwargs', 'dict(net_arch=[512, 512, 512], n_critics=2)'),
             ('replay_buffer_class', 'HerReplayBuffer'),
             ('replay_buffer_kwargs',
              "dict( online_sampling=True, goal_selection_strategy='future', "
              'n_sampled_goal=4, )'),
             ('tau', 0.05),
             ('normalize', False)])

Environment Arguments

{'render': True}