Toybox: A Suite of Environments for Experimental Evaluation of Deep Reinforcement Learning
This addresses the problem of limited experimental evaluation tools for deep RL researchers, though it is incremental as it builds on existing Atari environments.
The authors tackled the challenge of evaluating deep reinforcement learning by introducing TOYBOX, a suite of high-performance, open-source Atari environments designed for experimental analysis, enabling experiments and analyses that are impossible in other environments.
Evaluation of deep reinforcement learning (RL) is inherently challenging. In particular, learned policies are largely opaque, and hypotheses about the behavior of deep RL agents are difficult to test in black-box environments. Considerable effort has gone into addressing opacity, but almost no effort has been devoted to producing high quality environments for experimental evaluation of agent behavior. We present TOYBOX, a new high-performance, open-source* subset of Atari environments re-designed for the experimental evaluation of deep RL. We show that TOYBOX enables a wide range of experiments and analyses that are impossible in other environments. *https://kdl-umass.github.io/Toybox/