Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Alshemali, Safeyah Khaled, Bauer, Daniel, Marton, Yuval
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2410.15173
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918519609753600
author Alshemali, Safeyah Khaled
Bauer, Daniel
Marton, Yuval
author_facet Alshemali, Safeyah Khaled
Bauer, Daniel
Marton, Yuval
contents The thematic fit estimation task measures semantic arguments' compatibility with a given semantic role for a given predicate. We investigate if autoregressive LLMs have consistent, expressible knowledge of event arguments' thematic fit by experimenting with various prompt designs, manipulating input context, reasoning, and output forms. We set a new state-of-the-art on thematic fit benchmarks, but show that closed and open weight LLMs respond differently to our prompting strategies: Closed models achieve better scores overall and benefit from multi-step reasoning, but they perform worse at filtering out generated sentences incompatible with the given predicate, role, and argument. Our analysis shows that lemma tuple input and sentence input result in surprisingly different thematic fit score distributions.
format Preprint
id arxiv_https___arxiv_org_abs_2410_15173
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Uncovering Autoregressive LLM Knowledge of Thematic Fit in Event Representation
Alshemali, Safeyah Khaled
Bauer, Daniel
Marton, Yuval
Computation and Language
Artificial Intelligence
The thematic fit estimation task measures semantic arguments' compatibility with a given semantic role for a given predicate. We investigate if autoregressive LLMs have consistent, expressible knowledge of event arguments' thematic fit by experimenting with various prompt designs, manipulating input context, reasoning, and output forms. We set a new state-of-the-art on thematic fit benchmarks, but show that closed and open weight LLMs respond differently to our prompting strategies: Closed models achieve better scores overall and benefit from multi-step reasoning, but they perform worse at filtering out generated sentences incompatible with the given predicate, role, and argument. Our analysis shows that lemma tuple input and sentence input result in surprisingly different thematic fit score distributions.
title Uncovering Autoregressive LLM Knowledge of Thematic Fit in Event Representation
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2410.15173