On Learning to Summarize with Large Language Models as References

Yixin Liu, Kejian Shi, Katherine S He +5

Recent studies have found that summaries generated by large language models (LLMs) are favored by human annotators over the original reference summaries in commonly used summarizat…