Negative Mechanistic Interpretability Result : Strong mediators do not determine transitions between semantically salient states
Summary We present a project testing whether transformer based language models represent semantic and factual relations as simple vector offsets, fitted relation operators, structured rotation-like ma…
Read More


