SLM Lab

Modular Deep Reinforcement Learning framework in PyTorch.

SLM Lab is a software framework for reproducible reinforcement learning (RL) research. It enables easy development of RL algorithms using modular components and file-based configuration. It also enables flexible experimentation completed with hyperparameter search, result analysis and benchmark results.

📖 SLM Lab is also the companion library of the book Foundations of Deep Reinforcement Learning. The book's website and errata is here.

✨ Features

• Modular design for building deep RL algorithms

• Reproducibility using spec file and git SHA

• Experiment framework with automatic analysis

• Extensive benchmark results

• Well-tuned algorithm implementations

• Multiple RL environment offerings

Algorithms

SLM Lab implements most of the canonical RL algorithms:

• SARSA

• DQN (Deep Q-Network)

• Double-DQN, Dueling-DQN, PER (Prioritized Experience Replay)

• REINFORCE

• A2C (Advantage Actor-Critic) with GAE & n-step

• PPO (Proximal Policy Optimization)

• SAC (Soft Actor-Critic)

• SIL (Self Imitation Learning)

• Asynchronous version of all the above

They are implemented in a modular way such that differences in algorithm performance can be confidently ascribed to differences between algorithms, not between implementations.

Environments

SLM Lab currently includes the following environment offerings:

• OpenAI gym

• OpenAI Roboschool

• VizDoom (credit: joelouismarino)

• Unity environments with prebuilt binaries

Citation

If you use SLM Lab in your publication, please cite below:

@misc{kenggraesser2017slmlab,
    author = {Keng, Wah Loon and Graesser, Laura},
    title = {SLM Lab},
    year = {2017},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\url{https://github.com/kengz/SLM-Lab}},
}

License

This project is licensed under the MIT License.