Linear-time Minimum Bayes Risk Decoding with Reference Aggregation

Jannis Vamvas, Rico Sennrich
2/6/2024
cs.CL

Official code: ZurichNLP/mbr (61 stars, Apache-2.0, last commit 2024-06-03).

Abstract

Minimum Bayes Risk (MBR) decoding is a text generation technique that has been shown to improve the quality of machine translations, but is expensive, even if a sampling-based approximation is used. Besides requiring a large number of sampled sequences, it requires the pairwise calculation of a utility metric, which has quadratic complexity. In this paper, we propose to approximate pairwise metric scores with scores calculated against aggregated reference representations. This changes the complexity of utility estimation from $O(n^2)$ to $O(n)$, while empirically preserving most of the quality gains of MBR decoding. We release our source code at https://github.com/ZurichNLP/mbr

View on arXivView PDF

Code Implementations(1)

ZurichNLP/mbrOfficial100%
Apache-2.0

Minimum Bayes Risk Decoding for Hugging Face Transformers

617Shell, PythonOct 31, 20232 years agoApache-2.0

Similar Papers

Agreement-Constrained Probabilistic Minimum Bayes Risk Decoding

87% match2025

Efficient Minimum Bayes Risk Decoding using Low-Rank Matrix Completion Algorithms

85% match2026

mbrs: A Library for Minimum Bayes Risk Decoding

84% match1 repo532024

Structure-Conditional Minimum Bayes Risk Decoding

84% match1 repo2025

Mitigating Metric Bias in Minimum Bayes Risk Decoding

84% match2024

Hyperparameter-Free Approach for Faster Minimum Bayes Risk Decoding

83% match1 repo22024

Cite this paper

@article{vamvas2024lineartime,
  title  = {Linear-time Minimum Bayes Risk Decoding with Reference Aggregation},
  author = {Jannis Vamvas and Rico Sennrich},
  year   = {2024},
  eprint = {2402.04251},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2402.04251v2}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2402.04251.svg)](https://codeofpaper.com/paper/2402.04251)

Discussion