What is Colosseum?

\(\texttt{Colosseum}\) is a pioneering Python package that creates a bridge between theory and practice in tabular reinforcement learning with an eye on the non-tabular setting. Have a look at the Motivation section for a brief overview of the motivation behind the \(\texttt{Colosseum}\) project.

Getting Started

• Installation

• Quickstart

• Tutorials:

  • Configure \(\texttt{Colosseum}\)

  • Inspect Agents Performances

  • Inspect Markov Decision Processes

  • Visualize Markov Decision Processes

  • The \(\texttt{Colosseum}\) Benchmark

  • Analyse Benchmarking Results

  • Create Custom Benchmarks

  • Benchmarking Agents

  • Hyperparameters Optimization

  • Scale Benchmarking to a Cluster

  • Empirical Hardness Analysis

  • Non-Tabular Benchmarking

• API Documentation

• Contributions

• Discord channel

If you use \(\texttt{Colosseum}\) in your research, please cite the accompanying paper.

@inproceedings{conserva2022hardness,
  title={Hardness in Markov Decision Processes: Theory and Practice},
  author={Conserva, Michelangelo and Rauber, Paulo},
  year={2022},
  booktitle={Advances in Neural Information Processing Systems},
}

Acknowledgements
The authors would like to thank the open-source Python community for the fundamental tools this research has been built upon. In particular, the authors would like to thank the authors of \(\texttt{dm_env}\), Gin Config, Jupyter Notebook, Matplotlib, NetworkX, Numba, Numpy, OpenAi Gym, Pandas, Scipy, Seaborn, TensorFlow, and tqdm.