Interpretable Language Modeling via Induction-head Ngram Models

Open in new window