Geodesic
MMA: a fight for multilingual models acceleration
Zapiski Nauchnykh Seminarov POMI, Investigations on applied mathematics and informatics. Part IV, Tome 540 (2024), pp. 214-232

Voir la notice de l'article provenant de la source Math-Net.Ru

In this work we focus on common NLP model design: fine-tuning a multilingual language model with data for the target task in one language to solve this task in a different target language. We aim to determine how popular speedup techniques affect multilingual capabilities of Transformer-based model and additionally research the usage of this techniques in combination. As a result, we obtain the NERC model that can be effectively inferred on CPU and keeps multilingual properties across several test languages after being tuned and accelerated with only English data available. 
@article{ZNSL_2024_540_a11,
     author = {N. Sukhanovskii and M. Ryndin},
     title = {MMA: a fight for multilingual models acceleration},
     journal = {Zapiski Nauchnykh Seminarov POMI},
     pages = {214--232},
     publisher = {mathdoc},
     volume = {540},
     year = {2024},
     language = {en},
     url = {http://geodesic.mathdoc.fr/item/ZNSL_2024_540_a11/}
}
TY  - JOUR
AU  - N. Sukhanovskii
AU  - M. Ryndin
TI  - MMA: a fight for multilingual models acceleration
JO  - Zapiski Nauchnykh Seminarov POMI
PY  - 2024
SP  - 214
EP  - 232
VL  - 540
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/item/ZNSL_2024_540_a11/
LA  - en
ID  - ZNSL_2024_540_a11
ER  - 
%0 Journal Article
%A N. Sukhanovskii
%A M. Ryndin
%T MMA: a fight for multilingual models acceleration
%J Zapiski Nauchnykh Seminarov POMI
%D 2024
%P 214-232
%V 540
%I mathdoc
%U http://geodesic.mathdoc.fr/item/ZNSL_2024_540_a11/
%G en
%F ZNSL_2024_540_a11
N. Sukhanovskii; M. Ryndin. MMA: a fight for multilingual models acceleration. Zapiski Nauchnykh Seminarov POMI, Investigations on applied mathematics and informatics. Part IV, Tome 540 (2024), pp. 214-232. http://geodesic.mathdoc.fr/item/ZNSL_2024_540_a11/