Centroid-Based Efficient Minimum Bayes Risk Decoding

arXiv:2402.11197v1 Announce Type: new
Abstract: Minimum Bayes risk (MBR) decoding achieved state-of-the-art translation performance by using COMET, a neural metric that has a high correlation with human evaluation. However, MBR decoding requires quadratic time since it computes the expected score between a translation hypothesis and all reference translations. We propose centroid-based MBR (CBMBR) decoding to improve the speed of MBR decoding. Our method clusters the reference translations in the feature space, and then calculates the score using the centroids of …

abstract art arxiv bayes comet correlation cs.cl decoding evaluation human hypothesis performance reference risk state translation type