# Incremental Beam Manipulation for Natural Language Generation

arXiv: https://arxiv.org/abs/2102.02574 · HTML: https://mlsanity.com/paper/2102.02574/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 28 references that are in the map.
- First posted on arXiv 4 Feb 2021.

## Builds on (most cited first, 18 of 18)

- [The Curious Case of Neural Text Degeneration](https://mlsanity.com/paper/1904.09751/) (2019)
- [Six Challenges for Neural Machine Translation](https://mlsanity.com/paper/1706.03872/) (2017)
- [The E2E Dataset: New Challenges For End-to-End Generation](https://mlsanity.com/paper/1706.09254/) (2017)
- [Why We Need New Evaluation Metrics for NLG](https://mlsanity.com/paper/1707.06875/) (2017)
- [Beam Search Strategies for Neural Machine Translation](https://mlsanity.com/paper/1702.01806/) (2017)
- [Sequence-to-Sequence Learning as Beam-Search Optimization](https://mlsanity.com/paper/1606.02960/) (2016)
- [Correcting Length Bias in Neural Machine Translation](https://mlsanity.com/paper/1808.10006/) (2018)
- [On NMT Search Errors and Model Errors: Cat Got Your Tongue?](https://mlsanity.com/paper/1908.10090/) (2019)
- [Evaluating the State-of-the-Art of End-to-End Natural Language Generation: The E2E NLG Challenge](https://mlsanity.com/paper/1901.07931/) (2019)
- [Learning to Search Better Than Your Teacher](https://mlsanity.com/paper/1502.02206/) (2015)
- [Sequence-to-Sequence Generation for Spoken Dialogue via Deep Syntax Trees and Strings](https://mlsanity.com/paper/1606.05491/) (2016)
- [Trainable Greedy Decoding for Neural Machine Translation](https://mlsanity.com/paper/1702.02429/) (2017)
- [Stochastic Language Generation in Dialogue using Recurrent Neural Networks with Convolutional Sentence Reranking](https://mlsanity.com/paper/1508.01755/) (2015)
- [Leveraging Sentence Similarity in Natural Language Generation: Improving Beam Search using Range Voting](https://mlsanity.com/paper/1908.06288/) (2019)
- [A Fully Differentiable Beam Search Decoder](https://mlsanity.com/paper/1902.06022/) (2019)
- [A Continuous Relaxation of Beam Search for End-to-end Training of Neural Sequence Models](https://mlsanity.com/paper/1708.00111/) (2017)
- [Learning Beam Search Policies via Imitation Learning](https://mlsanity.com/paper/1811.00512/) (2018)
- [Char2char Generation with Reranking for the E2E NLG Challenge](https://mlsanity.com/paper/1811.05826/) (2018)

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

- [Improved Beam Search for Hallucination Mitigation in Abstractive Summarization](https://mlsanity.com/paper/2212.02712/) (2022)
- [GLUECons: A Generic Benchmark for Learning Under Constraints](https://mlsanity.com/paper/2302.10914/) (2023)
- [Looking for Confirmations: An Effective and Human-Like Visual Dialogue Strategy](https://mlsanity.com/paper/2109.05312/) (2021)
- [A Systematic Review of Data-to-Text NLG](https://mlsanity.com/paper/2402.08496/) (2024)
- [Medical Text Simplification: Optimizing for Readability with Unlikelihood Training and Reranked Beam Search Decoding](https://mlsanity.com/paper/2310.11191/) (2023)
- [On the Depth between Beam Search and Exhaustive Search for Text Generation](https://mlsanity.com/paper/2308.13696/) (2023)

## Abstract (arXiv)

The performance of natural language generation systems has improved substantially with modern neural networks. At test time they typically employ beam search to avoid locally optimal but globally suboptimal predictions. However, due to model errors, a larger beam size can lead to deteriorating performance according to the evaluation metric. For this reason, it is common to rerank the output of beam search, but this relies on beam search to produce a good set of hypotheses, which limits the potential gains. Other alternatives to beam search require changes to the training of the model, which restricts their applicability compared to beam search. This paper proposes incremental beam manipulation, i.e. reranking the hypotheses in the beam during decoding instead of only at the end. This way, hypotheses that are unlikely to lead to a good final output are discarded, and in their place hypotheses that would have been ignored will be considered instead. Applying incremental beam manipulation leads to an improvement of 1.93 and 5.82 BLEU points over vanilla beam search for the test sets of the E2E and WebNLG challenges respectively. The proposed method also outperformed a strong reranker by 1.04 BLEU points on the E2E challenge, while being on par with it on the WebNLG dataset.