OpenAI Retro Contest

April 5 to June 5, 2018

We're holding a transfer-learning contest using the Sonic The Hedgehog™ series of games for SEGA Genesis. In this contest, participants try to create the best agent for playing custom levels of the Sonic games — without having access to those levels during development. See our blog post for more details.

Here's how the contest works:

  1. Train or script your agent to play Sonic The Hedgehog™
  2. Submit your agent to us as a Docker container
  3. We evaluate your agent on a set of secret test levels
  4. Your agent's score appears on the leaderboard

This process is illustrated in the schematic below.

We believe that the next step for reinforcement learning is to leverage past experience to quickly learn new environments. Current algorithms are very prone to memorization and can't adapt well to new situations. While this contest focuses on video game levels, we hope the winning techniques will be applicable to a wide variety of domains.

The contest will run from April 5 to June 5 (2 months) and winners will receive some pretty cool trophies.

To see rules for the contest or to get started with it, look at the details page. To see the current best submissions check out the leaderboard. The blog post has an overview of the motivations for the contest and the tech report describes the benchmark used by the contest in detail.