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