Control-DAG: Constrained Decoding for Non-Autoregressive Directed Acyclic T5 using Weighted Finite State Automata

Jinghong Chen, Weizhe Lin, Jingbiao Mei, Bill Byrne


Abstract
The Directed Acyclic Transformer is a fast non-autoregressive (NAR) model that performs well in Neural Machine Translation. Two issues prevent its application to general Natural Language Generation (NLG) tasks: frequent Out-Of-Vocabulary (OOV) errors and the inability to faithfully generate entity names. We introduce Control-DAG, a constrained decoding algorithm for our Directed Acyclic T5 (DA-T5) model which offers lexical, vocabulary and length control. We show that Control-DAG significantly enhances DA-T5 on the Schema Guided Dialogue and the DART datasets, establishing strong NAR results for Task-Oriented Dialogue and Data-to-Text NLG.
Anthology ID:
2024.naacl-short.42
Volume:
Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 2: Short Papers)
Month:
June
Year:
2024
Address:
Mexico City, Mexico
Editors:
Kevin Duh, Helena Gomez, Steven Bethard
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
508–518
Language:
URL:
https://aclanthology.org/2024.naacl-short.42
DOI:
10.18653/v1/2024.naacl-short.42
Bibkey:
Cite (ACL):
Jinghong Chen, Weizhe Lin, Jingbiao Mei, and Bill Byrne. 2024. Control-DAG: Constrained Decoding for Non-Autoregressive Directed Acyclic T5 using Weighted Finite State Automata. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 2: Short Papers), pages 508–518, Mexico City, Mexico. Association for Computational Linguistics.
Cite (Informal):
Control-DAG: Constrained Decoding for Non-Autoregressive Directed Acyclic T5 using Weighted Finite State Automata (Chen et al., NAACL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.naacl-short.42.pdf