/rllib_tutorials

RLlib tutorials

Primary LanguageJupyter NotebookMIT LicenseMIT

RLlib Tutorials

These reinforcement learning tutorials use environments from OpenAI Gym to illustrate how to train policies in RLlib.

Getting Started

To get started use git to clone this public repository:

git clone https://github.com/DerwenAI/rllib_tutorials.git
cd rllib_tutorials

Then use pip to install the required dependencies:

python3 -m pip install -U pip
python3 -m pip install -r requirements.txt

Alternatively, if you use conda for installing Python packages:

conda create -n rllib_tutorials python=3.7
conda activate rllib_tutorials
python3 -m pip install -r requirements.txt

Use JupyterLab to run the notebooks. Connect into the directory for this repo, then launch JupyterLab with the command line:

jupyter-lab

Tutorial: Intro to Reinforcement Learning and Tour Through RLlib

Intro to Reinforcement Learning and Tour Through RLlib covers an introductory, hands-on coding tour through RLlib and related components of Ray used for reinforcement learning applications in Python. This webinar begins with a lecture that introduces reinforcement learning, including the essential concepts and terminology, plus show typical coding patterns used in RLlib. We'll also explore four different well-known reinforcement learning environments through hands-on coding. The intention is to compare and contrast across these environments to highlight the practices used in RLlib. Then we'll follow with Q&A.

Prerequisites

  • some Python programming experience
  • some familiarity with machine learning
  • clone/install the Git repo
  • no previous work in reinforcement learning
  • no previous hands-on experience with RLlib

Background

See also:

Tutorial: Using Reinforcement Learning: Custom Environments, Multi-Armed Bandits, Recommendation Systems

Using Reinforcement Learning begins with a brief tutorial about how to build custom Gym environments to use with RLlib, to use as a starting point. We’ll then explore hands-on coding for RL through two use cases:

  1. Contextual bandits with a financial portfolio optimization example–a real-world problem addressed with a “constrained” class of RL algorithms
  2. Building a recommender system with RLlib–new approaches to recommenders, which can be adapted to similar use cases

Prerequisites

  • Some Python programming experience
  • Some familiarity with machine learning
  • Clone/install the Git repo
  • Intro to Reinforcement Learning and Tour Through RLlib or equivalent

Resources

Ray Summit
June 22-24, 2021
online, free registration
https://www.anyscale.com/ray-summit-2021