|
Quick Software Introduction |
|
|
|
Page 5 of 10 What is this packageThe purpose of this package is to provide the software foundation for the reinforcement learning competition. This package contains all of the resources required to create agents and experiments and to test them on the testing versions of the competition problems.
Every experiment that is run using this software consists of four components: - An experiment program (a trainer) - An agent program - An environment program - The RL_glue communication software
We have taken care of RL_glue and the environments, so the important components for you are the agents and the trainers.
|