@markriedl.bsky.social Anyone look at some of the mechanistic interpretability ideas on open weight models for these exams with the
“Correct” solutions?