arXiv:2203.16804cs.CL
BRIO: Bringing Order to Abstractive Summarization
Early discovery. Linked by one identifiable commenter. This paper is outside the main ranking; a mention is not an endorsement.
Abstract
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.
This is why, whenever I can, I call RLHF/DPO "sequence level calibration" instead of "alignment tuning". Some precursors to RLHF: https://arxiv.org/abs/2210.00045 https://arxiv.org/abs/2203.16804
boodleboodleView on HNParent context