@inproceedings{park-etal-2025-decoding,
title = "Decoding Dense Embeddings: Sparse Autoencoders for Interpreting and Discretizing Dense Retrieval",
author = "Park, Seongwan and
Kim, Taeklim and
Ko, Youngjoong",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.emnlp-main.1345/",
doi = "10.18653/v1/2025.emnlp-main.1345",
pages = "26468--26485",
ISBN = "979-8-89176-332-6",
abstract = "Despite their strong performance, Dense Passage Retrieval (DPR) models suffer from a lackof interpretability. In this work, we propose a novel interpretability framework that leveragesSparse Autoencoders (SAEs) to decompose previously uninterpretable dense embeddings fromDPR models into distinct, interpretable latent concepts. We generate natural language descriptionsfor each latent concept, enabling human interpretations of both the dense embeddingsand the query-document similarity scores of DPR models. We further introduce Concept-Level Sparse Retrieval (CL-SR), a retrieval framework that directly utilizes the extractedlatent concepts as indexing units. CL-SR effectively combines the semantic expressiveness ofdense embeddings with the transparency and efficiency of sparse representations. We showthat CL-SR achieves high index-space and computational efficiency while maintaining robustperformance across vocabulary and semantic mismatches."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="park-etal-2025-decoding">
<titleInfo>
<title>Decoding Dense Embeddings: Sparse Autoencoders for Interpreting and Discretizing Dense Retrieval</title>
</titleInfo>
<name type="personal">
<namePart type="given">Seongwan</namePart>
<namePart type="family">Park</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Taeklim</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Youngjoong</namePart>
<namePart type="family">Ko</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-332-6</identifier>
</relatedItem>
<abstract>Despite their strong performance, Dense Passage Retrieval (DPR) models suffer from a lackof interpretability. In this work, we propose a novel interpretability framework that leveragesSparse Autoencoders (SAEs) to decompose previously uninterpretable dense embeddings fromDPR models into distinct, interpretable latent concepts. We generate natural language descriptionsfor each latent concept, enabling human interpretations of both the dense embeddingsand the query-document similarity scores of DPR models. We further introduce Concept-Level Sparse Retrieval (CL-SR), a retrieval framework that directly utilizes the extractedlatent concepts as indexing units. CL-SR effectively combines the semantic expressiveness ofdense embeddings with the transparency and efficiency of sparse representations. We showthat CL-SR achieves high index-space and computational efficiency while maintaining robustperformance across vocabulary and semantic mismatches.</abstract>
<identifier type="citekey">park-etal-2025-decoding</identifier>
<identifier type="doi">10.18653/v1/2025.emnlp-main.1345</identifier>
<location>
<url>https://aclanthology.org/2025.emnlp-main.1345/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>26468</start>
<end>26485</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Decoding Dense Embeddings: Sparse Autoencoders for Interpreting and Discretizing Dense Retrieval
%A Park, Seongwan
%A Kim, Taeklim
%A Ko, Youngjoong
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-332-6
%F park-etal-2025-decoding
%X Despite their strong performance, Dense Passage Retrieval (DPR) models suffer from a lackof interpretability. In this work, we propose a novel interpretability framework that leveragesSparse Autoencoders (SAEs) to decompose previously uninterpretable dense embeddings fromDPR models into distinct, interpretable latent concepts. We generate natural language descriptionsfor each latent concept, enabling human interpretations of both the dense embeddingsand the query-document similarity scores of DPR models. We further introduce Concept-Level Sparse Retrieval (CL-SR), a retrieval framework that directly utilizes the extractedlatent concepts as indexing units. CL-SR effectively combines the semantic expressiveness ofdense embeddings with the transparency and efficiency of sparse representations. We showthat CL-SR achieves high index-space and computational efficiency while maintaining robustperformance across vocabulary and semantic mismatches.
%R 10.18653/v1/2025.emnlp-main.1345
%U https://aclanthology.org/2025.emnlp-main.1345/
%U https://doi.org/10.18653/v1/2025.emnlp-main.1345
%P 26468-26485
Markdown (Informal)
[Decoding Dense Embeddings: Sparse Autoencoders for Interpreting and Discretizing Dense Retrieval](https://aclanthology.org/2025.emnlp-main.1345/) (Park et al., EMNLP 2025)
ACL