idea of the MAGICAL benchmark ("working on better IL benchmarks is a great idea", " sorely needed ", " a lot of 2

Neural Information Processing Systems 

We're glad that R3 likes the idea of the paper and believes that the design process and methods are sound.