Saved in:
Bibliographic Details
Main Authors: Xu, Liyan, Su, Zhenlin, Yu, Mo, Li, Jiangnan, Meng, Fandong, Zhou, Jie
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.08592
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909752574869504
author Xu, Liyan
Su, Zhenlin
Yu, Mo
Li, Jiangnan
Meng, Fandong
Zhou, Jie
author_facet Xu, Liyan
Su, Zhenlin
Yu, Mo
Li, Jiangnan
Meng, Fandong
Zhou, Jie
contents This work stems from an observed limitation of text encoders: embeddings may not be able to recognize fine-grained entities or events within encoded semantics, resulting in failed retrieval even in simple cases. To examine such behaviors, we first introduce a new evaluation dataset, CapRetrieval, in which passages are image captions and queries are phrases targeting entity or event concepts in diverse forms. Zero-shot evaluation suggests that encoders often struggle with these fine-grained matching, regardless of training sources or model size. Aiming for enhancement, we proceed to finetune encoders with our proposed data generation strategies, enabling a small 0.1B encoder to outperform the state-of-the-art 7B model. Within this process, we further uncover the granularity dilemma, a challenge for embeddings to capture fine-grained salience while aligning with overall semantics. Our dataset, code and models in this work are publicly released at https://github.com/lxucs/CapRetrieval.
format Preprint
id arxiv_https___arxiv_org_abs_2506_08592
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Dense Retrievers Can Fail on Simple Queries: Revealing The Granularity Dilemma of Embeddings
Xu, Liyan
Su, Zhenlin
Yu, Mo
Li, Jiangnan
Meng, Fandong
Zhou, Jie
Computation and Language
Machine Learning
This work stems from an observed limitation of text encoders: embeddings may not be able to recognize fine-grained entities or events within encoded semantics, resulting in failed retrieval even in simple cases. To examine such behaviors, we first introduce a new evaluation dataset, CapRetrieval, in which passages are image captions and queries are phrases targeting entity or event concepts in diverse forms. Zero-shot evaluation suggests that encoders often struggle with these fine-grained matching, regardless of training sources or model size. Aiming for enhancement, we proceed to finetune encoders with our proposed data generation strategies, enabling a small 0.1B encoder to outperform the state-of-the-art 7B model. Within this process, we further uncover the granularity dilemma, a challenge for embeddings to capture fine-grained salience while aligning with overall semantics. Our dataset, code and models in this work are publicly released at https://github.com/lxucs/CapRetrieval.
title Dense Retrievers Can Fail on Simple Queries: Revealing The Granularity Dilemma of Embeddings
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2506.08592