Parallel Iterative Edit Models for Local Sequence Transduction

Propose parallel iterative edit models for local sequence transduction. Introduce a framework consisting of two tasks: tagging, which chooses and arbitrarily reorders a subset of input tokens to keep, and insertion, which infills the missing tokens with another pretrained masked language model.

Data and Resources

Cite this as

Awasthi et al., Ng et al., Boyd, Rozovskaya and Roth (2024). Dataset: Parallel Iterative Edit Models for Local Sequence Transduction. https://doi.org/10.57702/9oxcqq51

DOI retrieved: December 16, 2024

Additional Info

Field Value
Created December 16, 2024
Last update December 16, 2024
Defined In https://doi.org/10.48550/arXiv.2111.09280
Author Awasthi et al.
More Authors
Ng et al.
Boyd
Rozovskaya and Roth
Homepage https://aclanthology.org/2019.emnlp-conference.123