AxonRLDemo

Quick-n-Dirty™ RL demo running against an OpenAI Gym server, built using:

• Elixir
• Nx
• Axon

Video of training run with CartPole v1 available.

Installation

If available in Hex, the package can be installed by adding axon_rl_demo to your list of dependencies in mix.exs:

def deps do
  [
    {:axon_rl_demo, "~> 0.1.0"}
  ]
end

Documentation can be generated with ExDoc and published on HexDocs. Once published, the docs can be found at https://hexdocs.pm/axon_rl_demo.

OpenAI Gym server

This requires an OpenAI Gym server to be running, using the Python3 package from the gym-http-server repo. The file requirements.py3 in this repo's root contains the pip dependencies to create and run the server. Using virtualenv:

python3 -m venv ENV
. ENV/bin/activate
pip install -r requirements.py3
gym-http-server

Note that some systems don't have Python3 installed as python3 so use whichever command you need to create a Python3 environment...

Once the server is up and running on your local (so you can see the agent playing CartPole-v1), run the Axon RL Demo agent!

Axon RL Demo

The usual suspects: clone, get deps, run:

mix deps.get
mix run -e "AxonRLDemo.run()"