Environment Generation for Zero-Shot Compositional Reinforcement Learning

Overview

This repository includes the implementation of `Environment Generation for Zero-Shot Compositional Reinforcement Learning' by Izzeddin Gur, Natasha Jaques, Yingjie Miao, Jongwook Choi,Manoj Tiwari, Honglak Lee, and Aleksandra Faust. NeurIPS'21.

There are two core components: (i) CoDE, Compositional Design of Environments, and (ii) the gMiniWoB framework. For now, it only hosts the main gMiniWoB implementation. The rest will be added soon.

Installation

• Download contents: svn export https://github.com/google-research/google-research/trunk/compositional_rl

Install Dependencies

Install Bootstrap

• Create a new directory: mkdir /path/to/compositional_rl/gwob/bootstrap/
• Download bootstrap files from https://github.com/twbs/bootstrap/releases/download/v4.3.1/bootstrap-4.3.1-dist.zip
• Extract /path/to/bootstrap/js/bootstrap.min.js and /path/to/bootstrap/css/bootstrap.min.css to /path/to/compositional_rl/gwob/bootstrap/.

Install MiniWoB

• Clone the MiniWoB project: git clone https://github.com/stanfordnlp/miniwob-plusplus
• Put the whole directory /path/to/miniwob-plusplus under /path/to/compositional_rl/gwob/

You should now have three folders /path/to/compositional_rl/gwob/gminiwob/, /path/to/compositional_rl/gwob/bootstrap/, and /path/to/compositional_rl/gwob/miniwob-plusplus/.

Generating random websites in gMiniWoB

• Open file:///path/to/compositional_rl/gwob/gminiwob/sample_random_website.html in a browser and click "START".
• Each time the "START" button is clicked, this will create a random gMiniWoB website using a subset of primitives available in gMiniWoB.