Towards Lossless Token Pruning in Late-Interaction Retrieval Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zong, Yuxuan, Piwowarski, Benjamin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910913135640576
author Zong, Yuxuan
Piwowarski, Benjamin
author_facet Zong, Yuxuan
Piwowarski, Benjamin
contents Late interaction neural IR models like ColBERT offer a competitive effectiveness-efficiency trade-off across many benchmarks. However, they require a huge memory space to store the contextual representation for all the document tokens. Some works have proposed using either heuristics or statistical-based techniques to prune tokens from each document. This however doesn't guarantee that the removed tokens have no impact on the retrieval score. Our work uses a principled approach to define how to prune tokens without impacting the score between a document and a query. We introduce three regularization losses, that induce a solution with high pruning ratios, as well as two pruning strategies. We study them experimentally (in and out-domain), showing that we can preserve ColBERT's performance while using only 30\% of the tokens.
format Preprint
id arxiv_https___arxiv_org_abs_2504_12778
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Towards Lossless Token Pruning in Late-Interaction Retrieval Models
Zong, Yuxuan
Piwowarski, Benjamin
Information Retrieval
Artificial Intelligence
Computation and Language
Late interaction neural IR models like ColBERT offer a competitive effectiveness-efficiency trade-off across many benchmarks. However, they require a huge memory space to store the contextual representation for all the document tokens. Some works have proposed using either heuristics or statistical-based techniques to prune tokens from each document. This however doesn't guarantee that the removed tokens have no impact on the retrieval score. Our work uses a principled approach to define how to prune tokens without impacting the score between a document and a query. We introduce three regularization losses, that induce a solution with high pruning ratios, as well as two pruning strategies. We study them experimentally (in and out-domain), showing that we can preserve ColBERT's performance while using only 30\% of the tokens.
title Towards Lossless Token Pruning in Late-Interaction Retrieval Models
topic Information Retrieval
Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2504.12778