lljll / faciledeepreinforcementlearning

A Facile guide for Deep Reinforcement Learning.

Chapter0 Get started

Choice of environment

Stable baselines is recommended as a improvement of OpenAI-baselines.

Setup

• Before starting, using Virtual Environment is highly recommended.
• Note: Baselines requires python3 (>=3.5)

Preparation of Virtual Environment (VENV) and Python3

Please refer the complete guide.

Here is an example using virtualenv and default interpreter python2.7 under Mac OSX

Install python3 by homebrew, which doesn't affet system python.

brew install python3

Create a virtual environmet named python3 under {your_folder}/virtualenv

virtualenv -p /usr/local/bin/python3 python3

activate the virtual environmet

. bin/activate

install needed packages under VENV Note: Stabe-Baselines supports Tensorflow versions from 1.8.0 to 1.14.0. Support for Tensorflow 2 API is planned.

sudo chown -R $(whoami) $(brew --prefix)/*
brew install cmake openmpi
pip install stable-baselines[mpi]`
pip install tensorflow

(optional) loading virtual environment in VSCode

If you hope to load virtual environment in VSCode editor, set in settings.json as below: "python.venvPath": "~/Codes/virtualenv"

Hello, world!

Chapter1

References

Essential Papers

Links

More resources

http://rail.eecs.berkeley.edu/deeprlcourse/