OpenAI's Automated Interpretability from paper "Language models can explain neurons in language models". Modified by Johnny Lin to add new models/context windows.
prepositions indicating spatial relationships or location, particularly "to", "at", "on", "in", and "of" when describing physical positioning or placement of objects.
variable assignments that track or store previous values or states, particularly for variables with names like "last", "old", or subscript patterns indicating previous time steps.
