Theoretical Guarantees for Minimum Bayes Risk Decoding
Not on the map. Why?
Abstract (arXiv)
Minimum Bayes Risk (MBR) decoding optimizes output selection by maximizing the expected utility value of an underlying human distribution. While prior work has shown the effectiveness of MBR decoding through empirical evaluation, few studies have analytically investigated why the method is effective. As a result of our analysis, we show that, given the size n of the reference hypothesis set used in computation, MBR decoding approaches the optimal solution with high probability at a rate of Oleft(n^-frac12right), under certain assumptions, even though the language space Y is significantly larger |Y|gg n. This result helps to theoretically explain the strong performance observed in several prior empirical studies on MBR decoding. In addition, we provide the performance gap for maximum-a-posteriori (MAP) decoding and compare it to MBR decoding. The result of this paper indicates that MBR decoding tends to converge to the optimal solution faster than MAP decoding in several cases.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 28 of its 45 references that are in the map. First posted on arXiv 18 Feb 2025.
Builds on
28 of 45 references.
References · 28
- 2022Training language models to follow instructions with human feedbackcited by 15,971
- 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
- 2019BERTScore: Evaluating Text Generation with BERTcited by 5,512
- 2020Learning to summarize from human feedbackcited by 2,588
- 2018Don't Give Me the Details, Just the Summary! Topic-Aware Convolutional Neural Networks for Extreme Summarizationcited by 1,317
- 2020COMET: A Neural Framework for MT Evaluationcited by 1,033
- 2018Computational Optimal Transportcited by 780
- 2017Sharp Minima Can Generalize For Deep Netscited by 466
- 2015A Neural Attention Model for Abstractive Sentence Summarizationcited by 396
- 2023xCOMET: Transparent Machine Translation Evaluation through Fine-grained Error Detectioncited by 265
- 2020Is MAP Decoding All You Need? The Inadequacy of the Mode in Neural Machine Translationcited by 138
- 2022Truncation Sampling as Language Model Desmoothingcited by 123
16 more
- 2021High Quality Rather than High Model Probability: Minimum Bayes Risk Decoding with Neural Metricscited by 104
- 2024MetricX-24: The Google Submission to the WMT 2024 Metrics Shared Taskcited by 93
- 2021Sampling-Based Approximations to Minimum Bayes Risk Decoding for Neural Machine Translationcited by 65
- 2023Epsilon Sampling Rocks: Investigating Sampling Strategies for Minimum Bayes Risk Decoding for Machine Translationcited by 58
- 2021Understanding the Properties of Minimum Bayes Risk Decoding in Neural Machine Translationcited by 56
- 2020Limitations of Autoregressive Models and Their Alternativescited by 55
- 2022Follow the Wisdom of the Crowd: Effective Text Generation via Minimum Bayes Risk Decodingcited by 54
- 2023It's MBR All the Way Down: Modern Generation Techniques Through the Lens of Minimum Bayes Riskcited by 51
- 2023Faster Minimum Bayes Risk Decoding with Confidence-based Pruningcited by 34
- 2024Linear-time Minimum Bayes Risk Decoding with Reference Aggregationcited by 21
- 2024Better Instruction-Following Through Minimum Bayes Riskcited by 19
- 2016Neural Machine Translation by Minimising the Bayes-risk with Respect to Syntactic Translation Latticescited by 17
- 2019Leveraging Sentence Similarity in Natural Language Generation: Improving Beam Search using Range Votingcited by 15
- 2024Hyperparameter-Free Approach for Faster Minimum Bayes Risk Decodingcited by 15
- 2024Improving Minimum Bayes Risk Decoding with Multi-Promptcited by 13
- 2022DC-MBR: Distributional Cooling for Minimum Bayesian Risk Decodingcited by 4
Built on by
Citing papers per year: 2025: 2; 2026: 1.
3 papers.