The AEA gym skill demonstrates how a custom Reinforcement Learning agent, that uses OpenAI's gym library, may be embedded into an AEA skill and connection.

Preparation instructions


Follow the Preliminaries and Installation sections from the AEA quick start.

Demo instructions

Create the agent

In the root directory, create the gym agent and enter the project.

aea create my_gym_agent
cd my_gym_agent

Add the gym skill

aea add skill fetchai/gym:0.1.0

Copy the gym environment to the agent directory

mkdir gyms
cp -a ../examples/gym_ex/gyms/. gyms/

Add a gym connection

aea add connection fetchai/gym:0.1.0

Update the connection config

aea config set connections.gym.config.env 'gyms.env.BanditNArmedRandom'

Install the skill dependencies

To install the gym package, a dependency of the gym skill, from Pypi run

aea install

Run the agent with the gym connection

aea run --connections gym

You will see the gym training logs.

AEA gym training logs

Delete the agent

When you're done, you can go up a level and delete the agent.

cd ..
aea delete my_gym_agent


This diagram shows the communication between the agent and the gym environment

sequenceDiagram participant Agent participant Environment activate Agent activate Environment Agent->>Environment: reset loop learn Agent->>Environment: act Environment->>Agent: percept end Agent->>Environment: close deactivate Agent deactivate Environment