Salvato in:
Dettagli Bibliografici
Autori principali: Zhu, Chencheng, Shimada, Kazutaka, Taniguchi, Tomoki, Ohkuma, Tomoko
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2412.20043
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915083711414272
author Zhu, Chencheng
Shimada, Kazutaka
Taniguchi, Tomoki
Ohkuma, Tomoko
author_facet Zhu, Chencheng
Shimada, Kazutaka
Taniguchi, Tomoki
Ohkuma, Tomoko
contents Large language models (LLMs) demonstrate the ability to learn in-context, offering a potential solution for scientific information extraction, which often contends with challenges such as insufficient training data and the high cost of annotation processes. Given that the selection of in-context examples can significantly impact performance, it is crucial to design a proper method to sample the efficient ones. In this paper, we propose STAYKATE, a static-dynamic hybrid selection method that combines the principles of representativeness sampling from active learning with the prevalent retrieval-based approach. The results across three domain-specific datasets indicate that STAYKATE outperforms both the traditional supervised methods and existing selection methods. The enhancement in performance is particularly pronounced for entity types that other methods pose challenges.
format Preprint
id arxiv_https___arxiv_org_abs_2412_20043
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle STAYKATE: Hybrid In-Context Example Selection Combining Representativeness Sampling and Retrieval-based Approach -- A Case Study on Science Domains
Zhu, Chencheng
Shimada, Kazutaka
Taniguchi, Tomoki
Ohkuma, Tomoko
Computation and Language
Large language models (LLMs) demonstrate the ability to learn in-context, offering a potential solution for scientific information extraction, which often contends with challenges such as insufficient training data and the high cost of annotation processes. Given that the selection of in-context examples can significantly impact performance, it is crucial to design a proper method to sample the efficient ones. In this paper, we propose STAYKATE, a static-dynamic hybrid selection method that combines the principles of representativeness sampling from active learning with the prevalent retrieval-based approach. The results across three domain-specific datasets indicate that STAYKATE outperforms both the traditional supervised methods and existing selection methods. The enhancement in performance is particularly pronounced for entity types that other methods pose challenges.
title STAYKATE: Hybrid In-Context Example Selection Combining Representativeness Sampling and Retrieval-based Approach -- A Case Study on Science Domains
topic Computation and Language
url https://arxiv.org/abs/2412.20043