Characterizing the Expressivity of Fixed-Precision Transformer Language Models

Open in new window