# Differentiable lower bound for expected BLEU score

arXiv: https://arxiv.org/abs/1712.04708 · HTML: https://mlsanity.com/paper/1712.04708/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Dec 2017.

## Built on by (most cited first, 6 of 6)

- [BOLT: Fast Energy-based Controlled Text Generation with Tunable Biases](https://mlsanity.com/paper/2305.12018/) (2023)
- [Don't Take It Literally: An Edit-Invariant Sequence Loss for Text Generation](https://mlsanity.com/paper/2106.15078/) (2021)
- [Model Fusion through Bayesian Optimization in Language Model Fine-Tuning](https://mlsanity.com/paper/2411.06710/) (2024)
- [Boosting Video Captioning with Dynamic Loss Network](https://mlsanity.com/paper/2107.11707/) (2021)
- [Differentiable N-gram Objective on Abstractive Summarization](https://mlsanity.com/paper/2202.04003/) (2022)
- [Machine Translation Models Stand Strong in the Face of Adversarial Attacks](https://mlsanity.com/paper/2309.06527/) (2023)

## Abstract (arXiv)

In natural language processing tasks performance of the models is often measured with some non-differentiable metric, such as BLEU score. To use efficient gradient-based methods for optimization, it is a common workaround to optimize some surrogate loss function. This approach is effective if optimization of such loss also results in improving target metric. The corresponding problem is referred to as loss-evaluation mismatch. In the present work we propose a method for calculation of differentiable lower bound of expected BLEU score that does not involve computationally expensive sampling procedure such as the one required when using REINFORCE rule from reinforcement learning (RL) framework.