Saved in:
Bibliographic Details
Main Authors: Porco, Aldo, Mehra, Dhruv, Malioutov, Igor, Radhakrishnan, Karthik, Keymanesh, Moniba, Preoţiuc-Pietro, Daniel, MacAvaney, Sean, Cheng, Pengxiang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.15070
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909617982799872
author Porco, Aldo
Mehra, Dhruv
Malioutov, Igor
Radhakrishnan, Karthik
Keymanesh, Moniba
Preoţiuc-Pietro, Daniel
MacAvaney, Sean
Cheng, Pengxiang
author_facet Porco, Aldo
Mehra, Dhruv
Malioutov, Igor
Radhakrishnan, Karthik
Keymanesh, Moniba
Preoţiuc-Pietro, Daniel
MacAvaney, Sean
Cheng, Pengxiang
contents Learned Sparse Retrieval (LSR) models encode text as weighted term vectors, which need to be sparse to leverage inverted index structures during retrieval. SPLADE, the most popular LSR model, uses FLOPS regularization to encourage vector sparsity during training. However, FLOPS regularization does not ensure sparsity among terms - only within a given query or document. Terms with very high Document Frequencies (DFs) substantially increase latency in production retrieval engines, such as Apache Solr, due to their lengthy posting lists. To address the issue of high DFs, we present a new variant of FLOPS regularization: DF-FLOPS. This new regularization technique penalizes the usage of high-DF terms, thereby shortening posting lists and reducing retrieval latency. Unlike other inference-time sparsification methods, such as stopword removal, DF-FLOPS regularization allows for the selective inclusion of high-frequency terms in cases where the terms are truly salient. We find that DF-FLOPS successfully reduces the prevalence of high-DF terms and lowers retrieval latency (around 10x faster) in a production-grade engine while maintaining effectiveness both in-domain (only a 2.2-point drop in MRR@10) and cross-domain (improved performance in 12 out of 13 tasks on which we tested). With retrieval latencies on par with BM25, this work provides an important step towards making LSR practical for deployment in production-grade search engines.
format Preprint
id arxiv_https___arxiv_org_abs_2505_15070
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle An Alternative to FLOPS Regularization to Effectively Productionize SPLADE-Doc
Porco, Aldo
Mehra, Dhruv
Malioutov, Igor
Radhakrishnan, Karthik
Keymanesh, Moniba
Preoţiuc-Pietro, Daniel
MacAvaney, Sean
Cheng, Pengxiang
Information Retrieval
Computation and Language
Learned Sparse Retrieval (LSR) models encode text as weighted term vectors, which need to be sparse to leverage inverted index structures during retrieval. SPLADE, the most popular LSR model, uses FLOPS regularization to encourage vector sparsity during training. However, FLOPS regularization does not ensure sparsity among terms - only within a given query or document. Terms with very high Document Frequencies (DFs) substantially increase latency in production retrieval engines, such as Apache Solr, due to their lengthy posting lists. To address the issue of high DFs, we present a new variant of FLOPS regularization: DF-FLOPS. This new regularization technique penalizes the usage of high-DF terms, thereby shortening posting lists and reducing retrieval latency. Unlike other inference-time sparsification methods, such as stopword removal, DF-FLOPS regularization allows for the selective inclusion of high-frequency terms in cases where the terms are truly salient. We find that DF-FLOPS successfully reduces the prevalence of high-DF terms and lowers retrieval latency (around 10x faster) in a production-grade engine while maintaining effectiveness both in-domain (only a 2.2-point drop in MRR@10) and cross-domain (improved performance in 12 out of 13 tasks on which we tested). With retrieval latencies on par with BM25, this work provides an important step towards making LSR practical for deployment in production-grade search engines.
title An Alternative to FLOPS Regularization to Effectively Productionize SPLADE-Doc
topic Information Retrieval
Computation and Language
url https://arxiv.org/abs/2505.15070