LinguAlchemy: Fusing Typological and Geographical Elements for Unseen Language Generalization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Adilazuarda, Muhammad Farid, Cahyawijaya, Samuel, Aji, Alham Fikri, Winata, Genta Indra, Purwarianti, Ayu
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916422554222592
author Adilazuarda, Muhammad Farid
Cahyawijaya, Samuel
Aji, Alham Fikri
Winata, Genta Indra
Purwarianti, Ayu
author_facet Adilazuarda, Muhammad Farid
Cahyawijaya, Samuel
Aji, Alham Fikri
Winata, Genta Indra
Purwarianti, Ayu
contents Pretrained language models (PLMs) have become remarkably adept at task and language generalization. Nonetheless, they often fail when faced with unseen languages. In this work, we present LinguAlchemy, a regularization method that incorporates various linguistic information covering typological, geographical, and phylogenetic features to align PLMs representation to the corresponding linguistic information on each language. Our LinguAlchemy significantly improves the performance of mBERT and XLM-R on low-resource languages in multiple downstream tasks such as intent classification, news classification, and semantic relatedness compared to fully finetuned models and displaying a high degree of unseen language generalization. We further introduce AlchemyScale and AlchemyTune, extension of LinguAlchemy which adjusts the linguistic regularization weights automatically, alleviating the need for hyperparameter search.
format Preprint
id arxiv_https___arxiv_org_abs_2401_06034
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LinguAlchemy: Fusing Typological and Geographical Elements for Unseen Language Generalization
Adilazuarda, Muhammad Farid
Cahyawijaya, Samuel
Aji, Alham Fikri
Winata, Genta Indra
Purwarianti, Ayu
Computation and Language
Pretrained language models (PLMs) have become remarkably adept at task and language generalization. Nonetheless, they often fail when faced with unseen languages. In this work, we present LinguAlchemy, a regularization method that incorporates various linguistic information covering typological, geographical, and phylogenetic features to align PLMs representation to the corresponding linguistic information on each language. Our LinguAlchemy significantly improves the performance of mBERT and XLM-R on low-resource languages in multiple downstream tasks such as intent classification, news classification, and semantic relatedness compared to fully finetuned models and displaying a high degree of unseen language generalization. We further introduce AlchemyScale and AlchemyTune, extension of LinguAlchemy which adjusts the linguistic regularization weights automatically, alleviating the need for hyperparameter search.
title LinguAlchemy: Fusing Typological and Geographical Elements for Unseen Language Generalization
topic Computation and Language
url https://arxiv.org/abs/2401.06034