1 |
Adapting BigScience Multilingual Model to Unseen Languages ...
|
|
|
|
BASE
|
|
Show details
|
|
2 |
Visualizing Cross-Lingual Discourse Relations in Multilingual TED Corpora
|
|
|
|
In: Proceedings of the 2nd Workshop on Computational Approaches to Discourse ; CODI 2021: 2nd Workshop on Computational Approaches to Discourse ; https://hal.archives-ouvertes.fr/hal-03642341 ; CODI 2021: 2nd Workshop on Computational Approaches to Discourse, Nov 2021, Punta Cana, Dominican Republic. ⟨10.18653/v1/2021.codi-main.16⟩ (2021)
|
|
BASE
|
|
Show details
|
|
3 |
Do Multilingual Neural Machine Translation Models Contain Language Pair Specific Attention Heads?
|
|
|
|
In: Findings of ACL 2021 ; https://hal.archives-ouvertes.fr/hal-03299010 ; Findings of ACL 2021, Aug 2021, Bangkok (virtual), Thailand (2021)
|
|
BASE
|
|
Show details
|
|
4 |
The Rediscovery Hypothesis: Language Models Need to Meet Linguistics ...
|
|
|
|
BASE
|
|
Show details
|
|
5 |
Do Multilingual Neural Machine Translation Models Contain Language Pair Specific Attention Heads? ...
|
|
|
|
BASE
|
|
Show details
|
|
6 |
Efficient Inference for Multilingual Neural Machine Translation ...
|
|
|
|
BASE
|
|
Show details
|
|
7 |
Multilingual Domain Adaptation for NMT: Decoupling Language and Domain Information with Adapters ...
|
|
|
|
Abstract:
Adapter layers are lightweight, learnable units inserted between transformer layers. Recent work explores using such layers for neural machine translation (NMT), to adapt pre-trained models to new domains or language pairs, training only a small set of parameters for each new setting (language pair or domain). In this work we study the compositionality of language and domain adapters in the context of Machine Translation. We aim to study, 1) parameter-efficient adaptation to multiple domains and languages simultaneously (full-resource scenario) and 2) cross-lingual transfer in domains where parallel data is unavailable for certain language pairs (partial-resource scenario). We find that in the partial resource scenario a naive combination of domain-specific and language-specific adapters often results in `catastrophic forgetting' of the missing languages. We study other ways to combine the adapters to alleviate this issue and maximize cross-lingual transfer. With our best adapter combinations, we obtain ... : Accepted at The Sixth Conference in Machine Translation (WMT21) ...
|
|
Keyword:
Computation and Language cs.CL; FOS Computer and information sciences
|
|
URL: https://arxiv.org/abs/2110.09574 https://dx.doi.org/10.48550/arxiv.2110.09574
|
|
BASE
|
|
Hide details
|
|
8 |
Do Multilingual Neural Machine Translation Models Contain Language Pair Specific Attention Heads? ...
|
|
|
|
BASE
|
|
Show details
|
|
9 |
Efficient Inference for Multilingual Neural Machine Translation ...
|
|
|
|
BASE
|
|
Show details
|
|
10 |
Multilingual Domain Adaptation for NMT: Decoupling Language and Domain Information with Adapters ...
|
|
|
|
BASE
|
|
Show details
|
|
11 |
A Multilingual Neural Machine Translation Model for Biomedical Data ...
|
|
|
|
BASE
|
|
Show details
|
|
12 |
A Syntax-Augmented Phrase-Based Statistical Machine Translation Model ; Modèle de traduction statistique à fragments enrichi par la syntaxe
|
|
|
|
In: https://tel.archives-ouvertes.fr/tel-00996317 ; Traitement du texte et du document. Université de Grenoble, 2010. Français (2010)
|
|
BASE
|
|
Show details
|
|
|
|