The idea was that we start building something like you have these huge parameter spaces, right? It's like excitatory strengths and hippotry strengths. The model would evolve over time. And so we built a tool called how abuse to start digging into that and kind of visualizing what effect does this parameters actually have on the evolution of the model. So you could drag a slider and see the excitatory strength of this to the model. And so this is how it evolves with time. That was really breakthrough to actually start analyzing. But it also meant that it eventually starts running more time building this visualization tool than what it's actually spending on my actual project.