BRIO：为抽象摘要带来秩序
BRIO: Bringing Order to Abstractive Summarization
摘要：
抽象摘要模型通常使用最大似然估计进行训练，该估计假设确定性（一点）目标分布，其中理想模型将所有概率质量分配给参考摘要。这种假设可能会导致推理过程中的性能下降，其中模型需要比较几个系统生成的（候选）摘要，这些摘要偏离了参考摘要。为了解决这个问题，我们提出了一种新的训练范式，它假设一个非确定性分布，使不同的候选摘要分配概率质量根据其质量。我们的方法在CNN/DailyMail（47.78 ROUGE-1）和XSum（49.07 ROUGE-1）数据集上实现了新的最先进的结果。进一步的分析还表明，我们的模型可以估计候选摘要的概率与其质量水平更相关。
英文摘要：
Abstractive summarization models are commonly trained using maximum likelihood estimation, which assumes a deterministic (one-point) target distribution in which an ideal model will assign all the probability mass to the reference summary. This assumption may lead to performance degradation during inference, where the model needs to compare several system-generated (candidate) summaries that have deviated from the reference summary. To address this problem, we propose a novel training paradigm which assumes a non-deterministic distribution so that different candidate summaries are assigned probability mass according to their quality. Our method achieves a new state-of-the-art result on the CNN/DailyMail (47.78 ROUGE-1) and XSum (49.07 ROUGE-1) datasets. Further analysis also shows that our model can estimate probabilities of candidate summaries that are more correlated with their level of quality.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
