# Modeling Confidence in Sequence-to-Sequence Models

arXiv: https://arxiv.org/abs/1910.01859 · HTML: https://mlsanity.com/paper/1910.01859/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 4 Oct 2019.

## Built on by (most cited first, 5 of 5)

- [Open Relation Modeling: Learning to Define Relations between Entities](https://mlsanity.com/paper/2108.09241/) (2021)
- [Perturbation-based QE: An Explainable, Unsupervised Word-level Quality Estimation Method for Blackbox Machine Translation](https://mlsanity.com/paper/2305.07457/) (2023)
- [How does a Language-Specific Tokenizer affect LLMs?](https://mlsanity.com/paper/2502.12560/) (2025)
- [Quality Estimation with k-nearest Neighbors and Automatic Evaluation for Model-specific Quality Estimation](https://mlsanity.com/paper/2404.18031/) (2024)
- [Using Source-Side Confidence Estimation for Reliable Translation into Unfamiliar Languages](https://mlsanity.com/paper/2503.23305/) (2025)

## Abstract (arXiv)

Recently, significant improvements have been achieved in various natural language processing tasks using neural sequence-to-sequence models. While aiming for the best generation quality is important, ultimately it is also necessary to develop models that can assess the quality of their output. In this work, we propose to use the similarity between training and test conditions as a measure for models' confidence. We investigate methods solely using the similarity as well as methods combining it with the posterior probability. While traditionally only target tokens are annotated with confidence measures, we also investigate methods to annotate source tokens with confidence. By learning an internal alignment model, we can significantly improve confidence projection over using state-of-the-art external alignment tools. We evaluate the proposed methods on downstream confidence estimation for machine translation (MT). We show improvements on segment-level confidence estimation as well as on confidence estimation for source tokens. In addition, we show that the same methods can also be applied to other tasks using sequence-to-sequence models. On the automatic speech recognition (ASR) task, we are able to find 60% of the errors by looking at 20% of the data.