Linguini: A benchmark for language-agnostic linguistic reasoning

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Sánchez, Eduardo, Alastruey, Belen, Ropers, Christophe, Stenetorp, Pontus, Artetxe, Mikel, Costa-jussà, Marta R.
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909319256080384
author Sánchez, Eduardo
Alastruey, Belen
Ropers, Christophe
Stenetorp, Pontus
Artetxe, Mikel
Costa-jussà, Marta R.
author_facet Sánchez, Eduardo
Alastruey, Belen
Ropers, Christophe
Stenetorp, Pontus
Artetxe, Mikel
Costa-jussà, Marta R.
contents We propose a new benchmark to measure a language model's linguistic reasoning skills without relying on pre-existing language-specific knowledge. The test covers 894 questions grouped in 160 problems across 75 (mostly) extremely low-resource languages, extracted from the International Linguistic Olympiad corpus. To attain high accuracy on this benchmark, models don't need previous knowledge of the tested language, as all the information needed to solve the linguistic puzzle is presented in the context. We find that, while all analyzed models rank below 25% accuracy, there is a significant gap between open and closed models, with the best-performing proprietary model at 24.05% and the best-performing open model at 8.84%.
format Preprint
id arxiv_https___arxiv_org_abs_2409_12126
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Linguini: A benchmark for language-agnostic linguistic reasoning
Sánchez, Eduardo
Alastruey, Belen
Ropers, Christophe
Stenetorp, Pontus
Artetxe, Mikel
Costa-jussà, Marta R.
Computation and Language
We propose a new benchmark to measure a language model's linguistic reasoning skills without relying on pre-existing language-specific knowledge. The test covers 894 questions grouped in 160 problems across 75 (mostly) extremely low-resource languages, extracted from the International Linguistic Olympiad corpus. To attain high accuracy on this benchmark, models don't need previous knowledge of the tested language, as all the information needed to solve the linguistic puzzle is presented in the context. We find that, while all analyzed models rank below 25% accuracy, there is a significant gap between open and closed models, with the best-performing proprietary model at 24.05% and the best-performing open model at 8.84%.
title Linguini: A benchmark for language-agnostic linguistic reasoning
topic Computation and Language
url https://arxiv.org/abs/2409.12126