/Reinforcement-Learning-2nd-Edition-by-Sutton-Codes

Working codes(exercises and in-chapter) for Reinforcement Learning, An Introduction by Richard Sutton

Primary LanguagePythonMIT LicenseMIT

Some Codes to work alongside Reinforcement Learning - An Introduction by Richard Sutton and Andrew G. Barto

Motivation

Detailed Solutions for each chapter an be found at Link - an excellent solution guide to help self-learners.

Just Found out such a repo exists already - Link Attached for reference

The aim of this repository is to collate some working codes hich shall be helpful in visualising the text. Some codes are written by me, others might be obtained from the internet, the source of which shall be referenced at the top of each code.

About Mistakes

This is a personal proect done while self-learning. Although the working state of each code is verified by me before updating the repository, some of these might have theoretical/conceptual/semantic errors, which I would be glad to correct when pointed out in issues.

Also incase someone wants to contribute additional code(or better code) are welcome to do so by raising a PR.

Usage

Other than OpenAI and Keras, other files should work on all versions of python. But Keras and OpenAI(owing to the use of tensorflow) require <=Python 3.8

Updates

  • TODO: Exercise 5.12: Designing the Racetrack is tedious so I am skipping this for now, but nonetheless this is a very interesting question(as we can create multiple tracks and a visualizer) and I shall attempt it at last
  • Exercise 7.2 is more of an empirical question rather than a programming question, hence is skipped.