dibyendu / Reinforcement-Learning

A playground for reinforcement learning algorithms

Standalone installation on local machine using python virtual environemnt

List the existing environment(s)

conda env list

Create a new conda environment

conda create -n rl python=3.8

Activate the environment

conda activate rl

Install the required packages

conda install pytorch cudatoolkit=10.2 -c pytorch
conda install -c conda-forge matplotlib==3.4.1
conda install -c conda-forge jupyter==1.0.0
conda install -c conda-forge pybox2d==2.3.10
conda install -c conda-forge seaborn==0.11.1
pip install jupyter_contrib_nbextensions==0.5.1
pip install PyVirtualDisplay==2.1
pip install mlagents-envs==0.25.1
pip install gym==0.18.0
pip install gym[atari]

Deactivate the environment

conda deactivate

(Optional) Export the environment

conda env export --no-builds -f environment.yml

(Optional) Delete the environment

conda remove --name rl --all

Docker installation

Build the container

docker build --tag image_name:image_tag github.com/dibyendu/Reinforcement-Learning#main

Run the container

docker run --publish 127.0.0.1:8080:5678 --publish 127.0.0.1:9999:9999 image_name:image_tag jupyter notebook --ip 0.0.0.0 --port 5678 --NotebookApp.custom_display_url=http://localhost:8080

(Optional) Delete the container

docker image rm --force image_name:image_tag