Revealing the structure of language model capabilities

Open in new window