mlsanityData to 13 Oct 2026

Breaking the Beam Search Curse: A Study of (Re-)Scoring Methods and Stopping Criteria for Neural Machine Translation

arXiv 1808.09582 ↗First posted 28 Aug 2018cs.CLcited by 29

Not on the map. Why?

Abstract (arXiv)

Beam search is widely used in neural machine translation, and usually improves translation quality compared to greedy search. It has been widely observed that, however, beam sizes larger than 5 hurt translation quality. We explain why this happens, and propose several methods to address this problem. Furthermore, we discuss the optimal stopping criteria for these methods. Results show that our hyperparameter-free methods outperform the widely-used hyperparameter-free heuristic of length normalization by +2.0 BLEU, and achieve the best results among all methods on Chinese-to-English translation.

Key facts. Cited by 29 papers in the map (data through 13 Oct 2026). First posted on arXiv 28 Aug 2018.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 13 Oct)

Citing papers per year: 2021: 8; 2022: 7; 2023: 6; 2024: 0; 2025: 1; 2026: 7.

29 papers.

Sort:

By area:

  1. 2022Calibrating Sequence likelihood Improves Conditional Language Generationcited by 121
  2. 2022Sequence-to-Sequence Knowledge Graph Completion and Question Answeringcited by 75
  3. 2022Quality-Aware Decoding for Neural Machine Translationcited by 49
  4. 2021Revisiting the Weaknesses of Reinforcement Learning for Neural Machine Translationcited by 32
  5. 2021Improving Multilingual Translation by Representation and Gradient Regularizationcited by 31
  6. 2021Searchable Hidden Intermediates for End-to-End Models of Decomposable Sequence Taskscited by 24
  7. 2021Reweighting Augmented Samples by Minimizing the Maximal Expected Losscited by 21
  8. 2023PathFinder: Guided Search over Multi-Step Reasoning Pathscited by 12
  9. 2021Smoothing and Shrinking the Sparse Seq2Seq Search Spacecited by 11
  10. 2022A Call for Clarity in Beam Search: How It Works and When It Stopscited by 10
  11. 2022Eliciting Knowledge from Large Pre-Trained Models for Unsupervised Knowledge-Grounded Conversationcited by 10
  12. 2023Look-back Decoding for Open-Ended Text Generationcited by 10
17 more
  1. 2021Multi-Sentence Resampling: A Simple Approach to Alleviate Dataset Length Bias and Beam-Search Degradationcited by 4
  2. 2022DC-MBR: Distributional Cooling for Minimum Bayesian Risk Decodingcited by 4
  3. 2023GEC-DePenD: Non-Autoregressive Grammatical Error Correction with Decoupled Permutation and Decodingcited by 3
  4. 2026When are likely answers right? On Sequence Probability and Correctness in LLMscited by 3
  5. 2021Searching for Search Errors in Neural Morphological Inflectioncited by 2
  6. 2021Digging Errors in NMT: Evaluating and Understanding Model Errors from Partial Hypothesis Spacecited by 2
  7. 2022The Implicit Length Bias of Label Smoothing on Beam Search Decodingcited by 2
  8. 2023On the Depth between Beam Search and Exhaustive Search for Text Generationcited by 2
  9. 2023Utility-Probability Duality of Neural Networkscited by 1
  10. 2025Increasing LLM response trustworthiness using voting ensemblescited by 1
  11. 2026Efficient Test-Time Inference via Deterministic Exploration of Truncated Decoding Treescited by 1
  12. 2023Bring More Attention to Syntactic Symmetry for Automatic Postediting of High-Quality Machine Translationscited by 0
  13. 2026Beyond Distribution Sharpening: The Importance of Task Rewardscited by 0
  14. 2026LBR: Towards Mitigating Length Bias in Large Language Models for Recommendationcited by 0
  15. 2026SpecCal: Ambiguity-Aware Candidate Calibration for Infrared Spectrum-Based Molecular Structure Reconstructioncited by 0
  16. 2026Beyond Tokens: A Survey on Decoding Methods for Large Language and Vision-Language Modelscited by 0
  17. 2026Beam Search, Self-Consistency, and the Limits of Inference-Time Scaling for Grammar-Constrained Text-to-SQL in Small Language Modelscited by 0