Artificial Curiosity

This repository contains the code and models for the Artificial Curiosity Visium webinar with Women++.

Open the Workshop in Google Colab

To run the notebook in Colab, click here ->

To see the output without running it, click here ->

Educational Resources & Bibliography

• OpenAI Gym toolkit to build environments

• Sutton and Barto, "Reinforcement Learning: An Introduction", MIT Press 2018'

• David Silver, Reinforcement Learning Course, UCL 2015: slides and videos

• Pathak et al., "Curiosity-driven Exploration by Self-supervised Prediction", 2017

• Burda, Edwards, Pathak et al., "Large-Scale Study of Curiosity-Driven Learning", 2018

• Savinov, Raichuk, Marinier, Vincent et al. "Episodic Curiosity through Reachability", 2018

How to run the code on your computer?

Build a Docker image from this repository

1. Install Docker
   • Windows 10, Mac OS, Linux: find your OS on https://docs.docker.com/install/#supported-platforms
   • Windows <10: follow the instructions on https://docs.docker.com/toolbox/toolbox_install_windows/ to install Docker Toolbox
2. Download this repository
3. Open a terminal and move in the repository
   • Windows 10, Mac OS, Linux: use your favorite terminal application
   • Windows <10: use the Docker Quickstart Terminal installed with Docker Toolbox
4. Build the Docker image: type docker build -t artificial_curiosity . (this might take a while)

Run the image

5. Still from the repository, run the Docker image:
   • Windows Command Line: type docker run -it -p 8888:8888 -v "%cd%":/app artificial_curiosity
   • Windows PowerShell / Mac OS / Linux: type docker run -it -p 8888:8888 -v ${PWD}:/app artificial_curiosity
6. The container will automatically start a Jupyter notebook server
   • Mac OS, Linux: go to http://localhost:8888
   • Windows: go to http://192.168.99.100:8888