Okay, Let's Do This! Modeling Event Coreference with Generated Rationales and Knowledge Distillation

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Nath, Abhijnan, Manafi, Shadi, Chelle, Avyakta, Krishnaswamy, Nikhil
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866929303034265600
author Nath, Abhijnan
Manafi, Shadi
Chelle, Avyakta
Krishnaswamy, Nikhil
author_facet Nath, Abhijnan
Manafi, Shadi
Chelle, Avyakta
Krishnaswamy, Nikhil
contents In NLP, Event Coreference Resolution (ECR) is the task of connecting event clusters that refer to the same underlying real-life event, usually via neural systems. In this work, we investigate using abductive free-text rationales (FTRs) generated by modern autoregressive LLMs as distant supervision of smaller student models for cross-document coreference (CDCR) of events. We implement novel rationale-oriented event clustering and knowledge distillation methods for event coreference scoring that leverage enriched information from the FTRs for improved CDCR without additional annotation or expensive document clustering. Our model using coreference specific knowledge distillation achieves SOTA B3 F1 on the ECB+ and GVC corpora and we establish a new baseline on the AIDA Phase 1 corpus. Our code can be found at https://github.com/csu-signal/llama_cdcr
format Preprint
id arxiv_https___arxiv_org_abs_2404_03196
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Okay, Let's Do This! Modeling Event Coreference with Generated Rationales and Knowledge Distillation
Nath, Abhijnan
Manafi, Shadi
Chelle, Avyakta
Krishnaswamy, Nikhil
Computation and Language
In NLP, Event Coreference Resolution (ECR) is the task of connecting event clusters that refer to the same underlying real-life event, usually via neural systems. In this work, we investigate using abductive free-text rationales (FTRs) generated by modern autoregressive LLMs as distant supervision of smaller student models for cross-document coreference (CDCR) of events. We implement novel rationale-oriented event clustering and knowledge distillation methods for event coreference scoring that leverage enriched information from the FTRs for improved CDCR without additional annotation or expensive document clustering. Our model using coreference specific knowledge distillation achieves SOTA B3 F1 on the ECB+ and GVC corpora and we establish a new baseline on the AIDA Phase 1 corpus. Our code can be found at https://github.com/csu-signal/llama_cdcr
title Okay, Let's Do This! Modeling Event Coreference with Generated Rationales and Knowledge Distillation
topic Computation and Language
url https://arxiv.org/abs/2404.03196