The need for understanding how models arrive at their predictions, and the techniques of mechanistic interpretability is within reach of anyone with a desire to learn.