FLOR: On the Effectiveness of Language Adaptation

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F25%3ACMIU2CIV" target="_blank" >RIV/00216208:11320/25:CMIU2CIV - isvavai.cz</a>

Výsledek na webu

<a href="https://aclanthology.org/2024.lrec-main.650" target="_blank" >https://aclanthology.org/2024.lrec-main.650</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

FLOR: On the Effectiveness of Language Adaptation

Popis výsledku v původním jazyce

Large language models have amply proven their great capabilities, both in downstream tasks and real-life settings. However, low- and mid-resource languages do not have access to the necessary means to train such models from scratch, and often have to rely on multilingual models despite being underrepresented in the training data. For the particular case of the Catalan language, we prove that continued pre-training with vocabulary adaptation is a better alternative to take the most out of already pre-trained models, even if these have not seen any Catalan data during their pre-training phase. We curate a 26B tokens corpus and use it to further pre-train BLOOM, giving rise to the FLOR models. We perform an extensive evaluation to assess the effectiveness of our method, obtaining consistent gains across Catalan and Spanish tasks. The models, training data, and evaluation framework are made freely available under permissive licenses.

Název v anglickém jazyce

FLOR: On the Effectiveness of Language Adaptation

Popis výsledku anglicky

Large language models have amply proven their great capabilities, both in downstream tasks and real-life settings. However, low- and mid-resource languages do not have access to the necessary means to train such models from scratch, and often have to rely on multilingual models despite being underrepresented in the training data. For the particular case of the Catalan language, we prove that continued pre-training with vocabulary adaptation is a better alternative to take the most out of already pre-trained models, even if these have not seen any Catalan data during their pre-training phase. We curate a 26B tokens corpus and use it to further pre-train BLOOM, giving rise to the FLOR models. We perform an extensive evaluation to assess the effectiveness of our method, obtaining consistent gains across Catalan and Spanish tasks. The models, training data, and evaluation framework are made freely available under permissive licenses.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

—

Návaznosti

—

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)

ISBN

978-2-493-81410-4

ISSN

—

e-ISSN

—

Počet stran výsledku

12

Strana od-do

7377-7388

Název nakladatele

ELRA and ICCL

Místo vydání

—

Místo konání akce

Torino, Italia

Datum konání akce

1. 1. 2025

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—