Affinity and Diversity: A Unified Metric for Demonstration Selection via Internal Representations

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Kato, Mariko, Cho, Hakaze, Sakai, Yoshihiro, Inoue, Naoya
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866913700406886400
author Kato, Mariko
Cho, Hakaze
Sakai, Yoshihiro
Inoue, Naoya
author_facet Kato, Mariko
Cho, Hakaze
Sakai, Yoshihiro
Inoue, Naoya
contents The performance of In-Context Learning (ICL) is highly sensitive to the selected demonstrations. Existing approaches to demonstration selection optimize different objectives, yielding inconsistent results. To address this, we propose a unified metric--affinity and diversity--that leverages ICL model's internal representations. Our experiments show that both affinity and diversity strongly correlate with test accuracies, indicating their effectiveness for demonstration selection. Moreover, we show that our proposed metrics align well with various previous works to unify the inconsistency.
format Preprint
id arxiv_https___arxiv_org_abs_2502_14380
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Affinity and Diversity: A Unified Metric for Demonstration Selection via Internal Representations
Kato, Mariko
Cho, Hakaze
Sakai, Yoshihiro
Inoue, Naoya
Computation and Language
Artificial Intelligence
Machine Learning
The performance of In-Context Learning (ICL) is highly sensitive to the selected demonstrations. Existing approaches to demonstration selection optimize different objectives, yielding inconsistent results. To address this, we propose a unified metric--affinity and diversity--that leverages ICL model's internal representations. Our experiments show that both affinity and diversity strongly correlate with test accuracies, indicating their effectiveness for demonstration selection. Moreover, we show that our proposed metrics align well with various previous works to unify the inconsistency.
title Affinity and Diversity: A Unified Metric for Demonstration Selection via Internal Representations
topic Computation and Language
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2502.14380