U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models

This paper investigates the rapidly advancing domain of Large Language Models (LLMs) and their growing potential in various fields. A central focus is the exploration of LLMs, e.g., LLaMA, as powerful tools for modeling and representing linguistic information, especially in the realm of syntax. We a...

Full description

Saved in:
Bibliographic Details
Main Authors: Claudiu Daniel Hromei, Danilo Croce, Roberto Basili
Format: Article
Language:English
Published: Accademia University Press 2024-07-01
Series:IJCoL
Online Access:https://journals.openedition.org/ijcol/1352
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850124186538737664
author Claudiu Daniel Hromei
Danilo Croce
Roberto Basili
author_facet Claudiu Daniel Hromei
Danilo Croce
Roberto Basili
author_sort Claudiu Daniel Hromei
collection DOAJ
description This paper investigates the rapidly advancing domain of Large Language Models (LLMs) and their growing potential in various fields. A central focus is the exploration of LLMs, e.g., LLaMA, as powerful tools for modeling and representing linguistic information, especially in the realm of syntax. We aim to evaluate the ability of these models to encode syntactic information, especially when explicitly supplied, through fine-tuning processes. Traditionally, Dependency Parsing has relied on specific techniques and dedicated architectures. Our research shifts this approach, conceptualizing it as a sequence-to-sequence task where Language Models interpret and transform syntax into bracketed structures that reflect dependency graphs. We introduce U-DepPLLaMA (Universal Dependency Parsing via auto-regressive LLMs based on LLaMA), a novel architecture optimized for multilingual, end-to-end Dependency Parsing. Our experimental evaluation, across 50 datasets in 26 languages from the Universal Dependency Treebank, shows that LLMs can be effectively trained for dependency parsing without the need for task-specific architectures. The results are on par with current state-of-the-art methods and demonstrate resilience across varying sentence complexities and lengths.
format Article
id doaj-art-332e8ece7ea747c1ac0cfedbb3084324
institution OA Journals
issn 2499-4553
language English
publishDate 2024-07-01
publisher Accademia University Press
record_format Article
series IJCoL
spelling doaj-art-332e8ece7ea747c1ac0cfedbb30843242025-08-20T02:34:24ZengAccademia University PressIJCoL2499-45532024-07-0110110.4000/125nmU-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language ModelsClaudiu Daniel HromeiDanilo CroceRoberto BasiliThis paper investigates the rapidly advancing domain of Large Language Models (LLMs) and their growing potential in various fields. A central focus is the exploration of LLMs, e.g., LLaMA, as powerful tools for modeling and representing linguistic information, especially in the realm of syntax. We aim to evaluate the ability of these models to encode syntactic information, especially when explicitly supplied, through fine-tuning processes. Traditionally, Dependency Parsing has relied on specific techniques and dedicated architectures. Our research shifts this approach, conceptualizing it as a sequence-to-sequence task where Language Models interpret and transform syntax into bracketed structures that reflect dependency graphs. We introduce U-DepPLLaMA (Universal Dependency Parsing via auto-regressive LLMs based on LLaMA), a novel architecture optimized for multilingual, end-to-end Dependency Parsing. Our experimental evaluation, across 50 datasets in 26 languages from the Universal Dependency Treebank, shows that LLMs can be effectively trained for dependency parsing without the need for task-specific architectures. The results are on par with current state-of-the-art methods and demonstrate resilience across varying sentence complexities and lengths.https://journals.openedition.org/ijcol/1352
spellingShingle Claudiu Daniel Hromei
Danilo Croce
Roberto Basili
U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models
IJCoL
title U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models
title_full U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models
title_fullStr U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models
title_full_unstemmed U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models
title_short U-DepPLLaMA: Universal Dependency Parsing via Auto-regressive Large Language Models
title_sort u deppllama universal dependency parsing via auto regressive large language models
url https://journals.openedition.org/ijcol/1352
work_keys_str_mv AT claudiudanielhromei udeppllamauniversaldependencyparsingviaautoregressivelargelanguagemodels
AT danilocroce udeppllamauniversaldependencyparsingviaautoregressivelargelanguagemodels
AT robertobasili udeppllamauniversaldependencyparsingviaautoregressivelargelanguagemodels