beobench 0.2.1

Beobench is a toolbox for benchmarking reinforcement learning (RL) algorithms on building energy optimisation (BEO) problems.

A toolbox for benchmarking reinforcement learning (RL) algorithms on building energy optimisation (BEO) problems. Beobench tries to make working on RL for BEO easier: it provides simple access to existing libraries defining BEO problems (such as BOPTEST) and provides a large set of pre-configured RL algorithms. Beobench is not a gym library itself - instead it leverages the brilliant work done by many existing gym-type projects and makes their work more easily accessible.

Features

Some of the features are work in progress

Main features

• RL algorithm collection: what’s the best RL method for your BEO problem? Building on Ray RLlib, beobench provides a large collection of pre-configured RL algorithm experiments that can be easily applied to your new BEO problem.

• Problem collection: beobench provides ready-to-use docker containers for popular BEO gym-type problem libraries. By enforcing a strict OpenAI gym.Env it makes testing your method on different libraries easy.

Additional features

• Experiment logging: log experiment results in a reproducible and shareable manner via Weights and Biases.

• Hyperparameter tuning: easily tune hyperparameters using the extensive Ray Tune Search API.

• Simple installation: beobench can be installed via pip and only requires docker as an additional non-python dependency.

• Easily extendable: beobench is designed for the user to add both environments and methods.

Quickstart

Run your first beobench experiment in three steps:

1. Install docker on your machine (if on Linux, check the additional installation steps)

2. Install beobench using:

   pip install beobench

3. Finally, start your first experiment using:

   python -m beobench.experiment.scheduler

Done, you have just started your first experiment… congrats! Check out the full getting started guide in the documentation for the next steps.

Documentation

https://beobench.readthedocs.io

License

MIT license

Credits

This package was originally created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.

History

0.2.1 (2022-02-03)

• Add integration with sinergym

• Move gym integrations to separate beobench_contrib repo

• Make usage of GPUs in containers optional

0.2.0 (2022-01-18)

• Enable adding custom environments to beobench with docker build context-based syntax

• Save experiment results on host machine

• Major improvements to documentation

• Remove unnecessary wandb arguments in main CLI

0.1.0 (2022-01-10)

• First release on PyPI.