jackoyoungblood/qrdqn-CartPole-v1

QRDQN Agent playing CartPole-v1

This is a trained model of a QRDQN agent playing CartPole-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

# Download model and save it into the logs/ folder
python -m utils.load_from_hub --algo qrdqn --env CartPole-v1 -orga jackoyoungblood -f logs/
python enjoy.py --algo qrdqn --env CartPole-v1  -f logs/

Training (with the RL Zoo)

python train.py --algo qrdqn --env CartPole-v1 -f logs/
# Upload the model and generate video (when possible)
python -m utils.push_to_hub --algo qrdqn --env CartPole-v1 -f logs/ -orga jackoyoungblood

Hyperparameters

OrderedDict([('batch_size', 64),
             ('buffer_size', 100000),
             ('exploration_final_eps', 0.04),
             ('exploration_fraction', 0.16),
             ('gamma', 0.99),
             ('gradient_steps', 128),
             ('learning_rate', 0.0023),
             ('learning_starts', 1000),
             ('n_timesteps', 50000.0),
             ('policy', 'MlpPolicy'),
             ('policy_kwargs', 'dict(net_arch=[256, 256], n_quantiles=10)'),
             ('target_update_interval', 10),
             ('train_freq', 256),
             ('normalize', False)])