Retrieving Examples from Memory for Retrieval Augmented Neural Machine Translation: A Systematic Comparison - Machine Learning and Information Access
Communication Dans Un Congrès Année : 2024

Retrieving Examples from Memory for Retrieval Augmented Neural Machine Translation: A Systematic Comparison

Une comparaison systématique des méthodes de pour la traduction automatique neuronale augmentée par des exemples

Maxime Bouthors
Josep Crego
  • Fonction : Auteur
François Yvon

Résumé

Retrieval-Augmented Neural Machine Translation (RAMT) architectures retrieve examples from memory to guide the generation process. While most works in this trend explore new ways to exploit the retrieved examples, the upstream retrieval step is mostly unexplored. In this paper, we study the effect of varying retrieval methods for several translation architectures to better understand the interplay between these two processes.We conduct experiments in two language pairs in a multi-domain setting and consider several downstream architectures based on a standard autoregressive model, an edit-based model, and a large language model with in-context learning. Our experiments show that the choice of the retrieval technique impacts the translation scores, with variance across architectures. We also discuss the effects of increasing the number and diversity of examples, which are mostly positive across the board.
Fichier principal
Vignette du fichier
2024.findings-naacl.190.pdf (398.26 Ko) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte

Dates et versions

hal-04670614 , version 1 (12-08-2024)

Licence

Identifiants

Citer

Maxime Bouthors, Josep Crego, François Yvon. Retrieving Examples from Memory for Retrieval Augmented Neural Machine Translation: A Systematic Comparison. 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL 2024), Association for Computational Linguistics, Jun 2024, Mexico, Mexico. pp.3022-3039, ⟨10.18653/v1/2024.findings-naacl.190⟩. ⟨hal-04670614⟩
49 Consultations
27 Téléchargements

Altmetric

Partager

More