In "Examining the Internals of the Model" it could be interesting to show an overview activation map that sums over the absolute activation of all the examples—

In "Examining the Internals of the Model" it could be interesting to show an overview activation map that sums over the absolute activation of all the examples—to highlight which cells activate on all examples—and one that is colored by variance of activation across examples—to show which cells influence the character/differences in writing.