A Warm Start and a Clean Crawled Corpus - A Recipe for Good Language Models

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F22%3AY9U39UU2" target="_blank" >RIV/00216208:11320/22:Y9U39UU2 - isvavai.cz</a>

Výsledek na webu

<a href="https://arxiv.org/pdf/2201.05601v2.pdf" target="_blank" >https://arxiv.org/pdf/2201.05601v2.pdf</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.48550/arXiv.2201.05601" target="_blank" >10.48550/arXiv.2201.05601</a>

Jazyk výsledku

angličtina

Název v původním jazyce

A Warm Start and a Clean Crawled Corpus - A Recipe for Good Language Models

Popis výsledku v původním jazyce

We train several language models for Icelandic, including IceBERT, that achieve state-of-the-art performance in a variety of downstream tasks, including part-of-speech tagging, named entity recognition, grammatical error detection and constituency parsing. To train the models we introduce a new corpus of Icelandic text, the Icelandic Common Crawl Corpus (IC3), a collection of high quality texts found online by targeting the Icelandic top-level-domain .is. Several other public data sources are also collected for a total of 16GB of Icelandic text. To enhance the evaluation of model performance and to raise the bar in baselines for Icelandic, we manually translate and adapt the WinoGrande commonsense reasoning dataset. Through these efforts we demonstrate that a properly cleaned crawled corpus is sufficient to achieve state-of-the-art results in NLP applications for low to medium resource languages, by comparison with models trained on a curated corpus. We further show that initializing models using existing multilingual models can lead to state-of-the-art results for some downstream tasks.

Název v anglickém jazyce

A Warm Start and a Clean Crawled Corpus - A Recipe for Good Language Models

Popis výsledku anglicky

We train several language models for Icelandic, including IceBERT, that achieve state-of-the-art performance in a variety of downstream tasks, including part-of-speech tagging, named entity recognition, grammatical error detection and constituency parsing. To train the models we introduce a new corpus of Icelandic text, the Icelandic Common Crawl Corpus (IC3), a collection of high quality texts found online by targeting the Icelandic top-level-domain .is. Several other public data sources are also collected for a total of 16GB of Icelandic text. To enhance the evaluation of model performance and to raise the bar in baselines for Icelandic, we manually translate and adapt the WinoGrande commonsense reasoning dataset. Through these efforts we demonstrate that a properly cleaned crawled corpus is sufficient to achieve state-of-the-art results in NLP applications for low to medium resource languages, by comparison with models trained on a curated corpus. We further show that initializing models using existing multilingual models can lead to state-of-the-art results for some downstream tasks.

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

2022

Kód důvěrnosti údajů

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

Název statě ve sborníku

Proceedings of the Thirteenth Language Resources and Evaluation Conference

ISBN

979-10-95546-72-6

ISSN

—

e-ISSN

—

Počet stran výsledku

11

Strana od-do

4356-4366

Název nakladatele

European Language Resources Association

Místo vydání

—

Místo konání akce

Marseille, France

Datum konání akce

1. 1. 2022

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—