/Reinforcement-Learning

An Open Source reimplementation of common RL algorithms.

Primary LanguagePythonApache License 2.0Apache-2.0

This library implements the core components of Reinforcement Learning agents. The goal is to have reusable components that can be composed flexibly together to tackle a wide range of scenarios.

Installation

Checkout this package and install it to your Python distribution.

$ git clone https://github.com/tomasruizt/Reinforcement-Learning.git
$ cd Reinforcement-Learning
$ pip install .

Usage

The library is structured in a hierarchy. You can import components from each hierachy and compose your Reinforcement Learning agents that way. For example:

from rl.policy.EpsilonGreedy import EpsilonGreedy
from rl.estimation_updater.MonteCarlo import MonteCarlo

mc = MonteCarlo()