User Guide

Note

This guide is for end users, not administrators. If you need help on the deployment process, please go to Deployment Guide.

aiVLE is short for AI Virtual Learning Environment. It is a platform for evaluating the performance of reinforcement learning algorithms. Since Python + OpenAI Gym is the most popular toolkit for both writing reinforcement learning environments and agents, aiVLE provides easy-to-use wrappers on top of existing OpenAI Gym environments and compatible agents for seamless migration from local/adhoc evaluation scripts to a online/reusable aiVLE Task.

To prepare an aiVLE Task, generally you need to provide two parts:

• (Optional) an aiVLE Gym environment: follow Creating Your Environment to convert an existing Gym
• An aiVLE Grader test suite: see Creating Your Test Suite

To validate your aiVLE Task before submitting to the platform, you may also want to take a look at Validating Sandbox to have a test-run of your entire package.