Standard
Same Neurons, Different Languages : Probing Morphosyntax in Multilingual Pre-trained Models. / Stańczak, Karolina; Ponti, Edoardo; Hennigen, Lucas Torroba; Cotterell, Ryan; Augenstein, Isabelle.
Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics (ACL), 2022. p. 1589-1598.
Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review
Harvard
Stańczak, K, Ponti, E, Hennigen, LT, Cotterell, R
& Augenstein, I 2022,
Same Neurons, Different Languages: Probing Morphosyntax in Multilingual Pre-trained Models. in
Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics (ACL), pp. 1589-1598, 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL 2022, Seattle, United States,
10/07/2022.
https://doi.org/10.18653/v1/2022.naacl-main.114
APA
Stańczak, K., Ponti, E., Hennigen, L. T., Cotterell, R.
, & Augenstein, I. (2022).
Same Neurons, Different Languages: Probing Morphosyntax in Multilingual Pre-trained Models. In
Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (pp. 1589-1598). Association for Computational Linguistics (ACL).
https://doi.org/10.18653/v1/2022.naacl-main.114
Vancouver
Stańczak K, Ponti E, Hennigen LT, Cotterell R
, Augenstein I.
Same Neurons, Different Languages: Probing Morphosyntax in Multilingual Pre-trained Models. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics (ACL). 2022. p. 1589-1598
https://doi.org/10.18653/v1/2022.naacl-main.114
Author
Stańczak, Karolina ; Ponti, Edoardo ; Hennigen, Lucas Torroba ; Cotterell, Ryan ; Augenstein, Isabelle. / Same Neurons, Different Languages : Probing Morphosyntax in Multilingual Pre-trained Models. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. Association for Computational Linguistics (ACL), 2022. pp. 1589-1598
Bibtex
@inproceedings{d6db1b5459904774b4f23795766642df,
title = "Same Neurons, Different Languages: Probing Morphosyntax in Multilingual Pre-trained Models",
abstract = "The success of multilingual pre-trained models is underpinned by their ability to learn representations shared by multiple languages even in absence of any explicit supervision. However, it remains unclear how these models learn to generalise across languages. In this work, we conjecture that multilingual pretrained models can derive language-universal abstractions about grammar. In particular, we investigate whether morphosyntactic information is encoded in the same subset of neurons in different languages. We conduct the first large-scale empirical study over 43 languages and 14 morphosyntactic categories with a state-of-the-art neuron-level probe. Our findings show that the cross-lingual overlap between neurons is significant, but its extent may vary across categories and depends on language proximity and pre-training data size.",
author = "Karolina Sta{\'n}czak and Edoardo Ponti and Hennigen, {Lucas Torroba} and Ryan Cotterell and Isabelle Augenstein",
note = "Publisher Copyright: {\textcopyright} 2022 Association for Computational Linguistics.; 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL 2022 ; Conference date: 10-07-2022 Through 15-07-2022",
year = "2022",
doi = "10.18653/v1/2022.naacl-main.114",
language = "English",
pages = "1589--1598",
booktitle = "Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
publisher = "Association for Computational Linguistics (ACL)",
address = "United States",
}
RIS
TY - GEN
T1 - Same Neurons, Different Languages
T2 - 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL 2022
AU - Stańczak, Karolina
AU - Ponti, Edoardo
AU - Hennigen, Lucas Torroba
AU - Cotterell, Ryan
AU - Augenstein, Isabelle
N1 - Publisher Copyright:
© 2022 Association for Computational Linguistics.
PY - 2022
Y1 - 2022
N2 - The success of multilingual pre-trained models is underpinned by their ability to learn representations shared by multiple languages even in absence of any explicit supervision. However, it remains unclear how these models learn to generalise across languages. In this work, we conjecture that multilingual pretrained models can derive language-universal abstractions about grammar. In particular, we investigate whether morphosyntactic information is encoded in the same subset of neurons in different languages. We conduct the first large-scale empirical study over 43 languages and 14 morphosyntactic categories with a state-of-the-art neuron-level probe. Our findings show that the cross-lingual overlap between neurons is significant, but its extent may vary across categories and depends on language proximity and pre-training data size.
AB - The success of multilingual pre-trained models is underpinned by their ability to learn representations shared by multiple languages even in absence of any explicit supervision. However, it remains unclear how these models learn to generalise across languages. In this work, we conjecture that multilingual pretrained models can derive language-universal abstractions about grammar. In particular, we investigate whether morphosyntactic information is encoded in the same subset of neurons in different languages. We conduct the first large-scale empirical study over 43 languages and 14 morphosyntactic categories with a state-of-the-art neuron-level probe. Our findings show that the cross-lingual overlap between neurons is significant, but its extent may vary across categories and depends on language proximity and pre-training data size.
UR - http://www.scopus.com/inward/record.url?scp=85138357220&partnerID=8YFLogxK
U2 - 10.18653/v1/2022.naacl-main.114
DO - 10.18653/v1/2022.naacl-main.114
M3 - Article in proceedings
AN - SCOPUS:85138357220
SP - 1589
EP - 1598
BT - Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies
PB - Association for Computational Linguistics (ACL)
Y2 - 10 July 2022 through 15 July 2022
ER -