Emma
Summary:
-
Large language models can improve their performance by proposing and debating individual responses over multiple rounds.
-
This 'society of minds' approach enhances mathematical and strategic reasoning while reducing fallacious answers and hallucinations.
Tags:
Research