Distribution Prompting: Understanding the Expressivity of Language Models Through the Next-Token Distributions They Can Produce

Open in new window