Learning to summarize with human feedback

NeurIPS 2020