afcarl / openai-k-armed-bandits

OpenAI Gym Environment for the k Armed Bandits Reinforcement Learning Task

openai-k-armed-bandits

This repository contains an easy-to-use OpenAI Gym environment to train your agents with the famously known k-Armed-Bandits Reinforcement Learning task. You can check out a working example of an agent using this environment in my Reinforcement Learning Book Exercices Repository..

Install

There are two ways to install this project. The easiest procedure is using pip to download the project from PyPi by typing in your terminal:

pip install gym_armed_bandits

An alternative way is by downloading the repository directly and installing the package locally:

git clone gym_armed_bandits
cd gym_armed_bandits
pip install -e .

This second method allows you to develop on the repository and instantly see the changes when importing. This is useful when contributing to this repository.

Use

In your Python scripts the libraries

import gym
import gym_armed_bandits

and then load a specific environment by running

env = gym.make(<environment name>)

e.g.

env = gym.make('ten-armed-bandits-v0')

Currently the existing environments available in this project are:

• three-armed-bandits-v0
• five-armed-bandits-v0
• ten-armed-bandits-v0

All of these environments have bandits with std=1 gaussian distributions.

The code is implemented in such a way that it is easy to add a new bandit.

License

MIT