Methods › Natural Language Processing › Autoregressive Transformers › Levenshtein Transformer

Levenshtein Transformer

12 papers tagged archive 2025-07-28

Introduced by Jiatao Gu et al. in Levenshtein Transformer

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

The Levenshtein Transformer (LevT) is a type of transformer that aims to address the lack of flexibility of previous decoding models. Notably, in previous frameworks, the length of generated sequences is either fixed or monotonically increased as the decoding proceeds. The authors argue this is incompatible with human-level intelligence where humans can revise, replace, revoke or delete any part of their generated text. Hence, LevT is proposed to bridge this gap by breaking the in-so-far standardized decoding mechanism and replacing it with two basic operations — insertion and deletion.

LevT is trained using imitation learning. The resulted model contains two policies and they are executed in an alternate manner. The authors argue that with this model decoding becomes more flexible. For example, when the decoder is given an empty token, it falls back to a normal sequence generation model. On the other hand, the decoder acts as a refinement model when the initial state is a low-quality generated sequence.

One crucial component in LevT framework is the learning algorithm. The authors leverage the characteristics of insertion and deletion — they are complementary but also adversarial. The algorithm they propose is called “dual policy learning”. The idea is that when training one policy (insertion or deletion), we use the output from its adversary at the previous iteration as input. An expert policy, on the other hand, is drawn to provide a correction signal.

PaperSourceSee Code · pytorch/fairseq

Papers archive 2025-07-28

12 shown of 12, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.

Tasks archive 2025-07-28

17 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Translation11
Machine Translation8
Imitation Learning3
Knowledge Distillation3
Diversity2
NMT2
Automatic Post-Editing1
Data Augmentation1
Domain Adaptation1
Optical Character Recognition (OCR)1
Reinforcement Learning (RL)1
Retrieval1
Scene Text Recognition1
Task 21
Text Summarization1
Transfer Learning1
reinforcement-learning1

Usage over time archive 2025-07-28

Papers per year tagged with Levenshtein Transformer: 2019 to 2024, peak 4 4 0 2019: 1 paper 2019 2020: 3 papers 2020 2021: 4 papers 2021 2022: 2 papers 2022 2023: 1 paper 2023 2024: 1 paper 2024
Papers per year the archive tags with this method, by the paper's archive date (12 dated). Bars are counts, not a trend claim.

Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).

Categories archive 2025-07-28

Autoregressive TransformersTransformers

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections