I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures

Timothee Mickus; Raul Vazquez; Joseph Attieh

I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures

Timothee Mickus, Raul Vazquez, Joseph Attieh

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

Abstract

Modularity is a paradigm of machine translation with the potential of bringing forth models that are large at training time and small during inference. Within this field of study, modular approaches, and in particular attention bridges, have been argued to improve the generalization capabilities of models by fostering language-independent representations. In the present paper, we study whether modularity affects translation quality; as well as how well modular architectures generalize across different evaluation scenarios. For a given computational budget, we find non-modular architectures to be always comparable or preferable to all modular designs we study.

Original language	English
Title of host publication	Proceedings of the Fifth Workshop on Insights from Negative Results in NLP
Editors	Shabnam Tafreshi, Arjun Akula, João Sedoc, Aleksandr Drozd, Anna Rogers, Anna Rumshisky
Number of pages	7
Place of Publication	Kerrville
Publisher	The Association for Computational Linguistics
Publication date	1 Jun 2024
Pages	34-40
ISBN (Electronic)	979-8-89176-102-5
Publication status	Published - 1 Jun 2024
MoE publication type	A4 Article in conference proceedings
Event	Workshop on Insights from Negative Results in NLP - Mexico City, Mexico Duration: 20 Jun 2024 → 20 Jun 2024 Conference number: 5

Fields of Science

6121 Languages
113 Computer and information sciences

Access to Document

2024.insights-1.5Final published version, 506 KBLicence: CC BY

https://aclanthology.org/2024.insights-1.5Licence: CC BY

FoTran: Found in Translation - Natural Language Understanding with Cross-Lingual Grounding
Tiedemann, J. (Project manager), Attieh, J. (Participant), Aulamo, M. (Participant), Celikkanat, H. (Participant), De Gibert Bonet, O. (Participant), Grönroos, S.-A. (Participant), Mickus, T. (Participant), Scherrer, Y. (Participant), Silfverberg, M. (Participant), Talman, A. (Participant), Vazquez , R. (Participant), Virpioja, S. P. (Participant), Yli-Jyrä, A. (Participant), Celikkanat, H. (Participant), Raganato, A. (Participant), Silfverberg, M. (Participant), Sulubacak, U. (Participant) & Vazquez , R. (Participant)
01/09/2018 → 31/03/2024
Project: The European Research Council: Consolidator Grant (H2020-ERC-COG)

Best paper Award
Varjonen, S. (Recipient), 7 Dec 2011
Prize: Prizes and awards

Cite this

Mickus, T., Vazquez, R., & Attieh, J. (2024). I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures. In S. Tafreshi, A. Akula, J. Sedoc, A. Drozd, A. Rogers, & A. Rumshisky (Eds.), Proceedings of the Fifth Workshop on Insights from Negative Results in NLP (pp. 34-40). The Association for Computational Linguistics. https://aclanthology.org/2024.insights-1.5

Mickus, Timothee ; Vazquez, Raul ; Attieh, Joseph. / I Have an Attention Bridge to Sell You : Generalization Capabilities of Modular Translation Architectures. Proceedings of the Fifth Workshop on Insights from Negative Results in NLP. editor / Shabnam Tafreshi ; Arjun Akula ; João Sedoc ; Aleksandr Drozd ; Anna Rogers ; Anna Rumshisky. Kerrville : The Association for Computational Linguistics, 2024. pp. 34-40

@inproceedings{eb7699c9fd5f4c8086e414179b7ec8cf,

title = "I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures",

abstract = "Modularity is a paradigm of machine translation with the potential of bringing forth models that are large at training time and small during inference. Within this field of study, modular approaches, and in particular attention bridges, have been argued to improve the generalization capabilities of models by fostering language-independent representations. In the present paper, we study whether modularity affects translation quality; as well as how well modular architectures generalize across different evaluation scenarios. For a given computational budget, we find non-modular architectures to be always comparable or preferable to all modular designs we study.",

keywords = "6121 Languages, 113 Computer and information sciences",

author = "Timothee Mickus and Raul Vazquez and Joseph Attieh",

year = "2024",

month = jun,

day = "1",

language = "English",

pages = "34--40",

editor = "Shabnam Tafreshi and Arjun Akula and Jo{\~a}o Sedoc and Aleksandr Drozd and Anna Rogers and Anna Rumshisky",

booktitle = "Proceedings of the Fifth Workshop on Insights from Negative Results in NLP",

publisher = "The Association for Computational Linguistics",

address = "United States",

note = "Workshop on Insights from Negative Results in NLP ; Conference date: 20-06-2024 Through 20-06-2024",

}

Mickus, T , Vazquez, R & Attieh, J 2024, I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures. in S Tafreshi, A Akula, J Sedoc, A Drozd, A Rogers & A Rumshisky (eds), Proceedings of the Fifth Workshop on Insights from Negative Results in NLP. The Association for Computational Linguistics, Kerrville, pp. 34-40, Workshop on Insights from Negative Results in NLP, Mexico City, Mexico, 20/06/2024. <https://aclanthology.org/2024.insights-1.5>

I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures. / Mickus, Timothee ; Vazquez, Raul ; Attieh, Joseph.
Proceedings of the Fifth Workshop on Insights from Negative Results in NLP. ed. / Shabnam Tafreshi; Arjun Akula; João Sedoc; Aleksandr Drozd; Anna Rogers; Anna Rumshisky. Kerrville: The Association for Computational Linguistics, 2024. p. 34-40.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - I Have an Attention Bridge to Sell You

T2 - Workshop on Insights from Negative Results in NLP

AU - Mickus, Timothee

AU - Vazquez, Raul

AU - Attieh, Joseph

N1 - Conference code: 5

PY - 2024/6/1

Y1 - 2024/6/1

N2 - Modularity is a paradigm of machine translation with the potential of bringing forth models that are large at training time and small during inference. Within this field of study, modular approaches, and in particular attention bridges, have been argued to improve the generalization capabilities of models by fostering language-independent representations. In the present paper, we study whether modularity affects translation quality; as well as how well modular architectures generalize across different evaluation scenarios. For a given computational budget, we find non-modular architectures to be always comparable or preferable to all modular designs we study.

AB - Modularity is a paradigm of machine translation with the potential of bringing forth models that are large at training time and small during inference. Within this field of study, modular approaches, and in particular attention bridges, have been argued to improve the generalization capabilities of models by fostering language-independent representations. In the present paper, we study whether modularity affects translation quality; as well as how well modular architectures generalize across different evaluation scenarios. For a given computational budget, we find non-modular architectures to be always comparable or preferable to all modular designs we study.

KW - 6121 Languages

KW - 113 Computer and information sciences

M3 - Conference contribution

SP - 34

EP - 40

BT - Proceedings of the Fifth Workshop on Insights from Negative Results in NLP

A2 - Tafreshi, Shabnam

A2 - Akula, Arjun

A2 - Sedoc, João

A2 - Drozd, Aleksandr

A2 - Rogers, Anna

A2 - Rumshisky, Anna

PB - The Association for Computational Linguistics

CY - Kerrville

Y2 - 20 June 2024 through 20 June 2024

ER -

I Have an Attention Bridge to Sell You: Generalization Capabilities of Modular Translation Architectures

Abstract

Fields of Science

Access to Document

Projects

FoTran: Found in Translation - Natural Language Understanding with Cross-Lingual Grounding

Prizes

Best paper Award

Cite this