ReAgent is an open source end-to-end platform for applied reinforcement learning (RL) developed and used at Facebook. ReAgent is built in Python and uses PyTorch for modeling and training and TorchScript for model serving. The platform contains workflows to train popular deep RL algorithms and includes data preprocessing, feature transformation, distributed training, counterfactual policy evaluation, and optimized serving. For more detailed information about ReAgent see the white paper here.
The platform was once named "Horizon" but we have adopted the name "ReAgent" recently to emphasize its broader scope in decision making and reasoning.
- Discrete-Action DQN
- Parametric-Action DQN
- Double DQN, Dueling DQN, Dueling Double DQN
- Distributional RL: C51 and QR-DQN
- Twin Delayed DDPG (TD3)
- Soft Actor-Critic (SAC)
ReAgent can be installed via. Docker or manually. Detailed instructions on how to install ReAgent can be found here.
Detailed instructions on how to use ReAgent Models can be found here.
The ReAgent Serving Platform (RASP) tutorial is available here.
ReAgent is released under a BSD 3-Clause license. Find out more about it here.
@article{gauci2018horizon, title={Horizon: Facebook's Open Source Applied Reinforcement Learning Platform}, author={Gauci, Jason and Conti, Edoardo and Liang, Yitao and Virochsiri, Kittipat and Chen, Zhengxing and He, Yuchen and Kaden, Zachary and Narayanan, Vivek and Ye, Xiaohui}, journal={arXiv preprint arXiv:1811.00260}, year={2018} }