return to table of content

Tracing the thoughts of a large language model