Mono- and cross-lingual evaluation of representation language models on less-resourced languages

Ulčar, Matej (avtor), Žagar, Aleš (avtor), Armendariz, Carlos S. (avtor), Repar, Andraž (avtor), Pollak, Senja (avtor), Purver, Matthew (avtor), Robnik Šikonja, Marko (avtor)

The current dominance of large language models in natural language processing is based on their contextual awareness. For text classification, text representation models, such as ELMo, BERT, and BERT derivatives, are typically fine-tuned for a specific problem. Most existing work focuses on English; in contrast, we present a large-scale multilingual empirical comparison of several monolingual and multilingual ELMo and BERT models using 14 classification tasks in nine languages. The results show, that the choice of best model largely depends on the task and language used, especially in a cross-lingual setting. In monolingual settings, monolingual BERT models tend to perform the best among BERT models. Among ELMo models, the ones trained on large corpora dominate. Cross-lingual knowledge transfer is feasible on most tasks already in a zero-shot setting without losing much performance.

• Jezik: Angleški jezik
• Ključne besede: monolingual models, multilingual models, ELMo, BERT, corpus, cross-lingual datasets, language models, contextual embeddings, less-resourced languages, BERT, ELMo
• Vrsta gradiva: Članek v reviji
• Tipologija: 1.01 - Izvirni znanstveni članek
• Organizacija: FRI - Fakulteta za računalništvo in informatiko
• Status publikacije: Objavljeno
• Različica publikacije: Objavljena publikacija
• Leto izida: 2026
• Št. strani: 29 str.
• Številčenje: Vol. 95, art. 101852
• PID: 20.500.12556/RUL-182550
• UDK: 004.8
• ISSN pri članku: 1095-8363
• DOI: 10.1016/j.csl.2025.101852
• COBISS.SI-ID: 241622275
• Datum objave v RUL: 15.05.2026

Gradivo je del revije

• Naslov: Computer speech & language
• Skrajšan naslov: Comput. speech lang.
• Založnik: Elsevier
• ISSN: 1095-8363
• COBISS.SI-ID: 203927043

Licence

• Licenca: CC BY 4.0, Creative Commons Priznanje avtorstva 4.0 Mednarodna
• Povezava: http://creativecommons.org/licenses/by/4.0/deed.sl
• Opis: To je standardna licenca Creative Commons, ki daje uporabnikom največ možnosti za nadaljnjo uporabo dela, pri čemer morajo navesti avtorja.

Sekundarni jezik

• Jezik: Slovenski jezik
• Ključne besede: korpusi, večjezični veliki modeli