Saved in:
Bibliographic Details
Main Authors: Gu, Geonmo, Heo, Byeongho, Yu, Jaemyung, Hwang, Jaehui, Kim, Taekyung, Lee, Sangmin, Jun, HeeJae, Kang, Yoohoon, Yun, Sangdoo, Han, Dongyoon
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2602.06393
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917379784572928
author Gu, Geonmo
Heo, Byeongho
Yu, Jaemyung
Hwang, Jaehui
Kim, Taekyung
Lee, Sangmin
Jun, HeeJae
Kang, Yoohoon
Yun, Sangdoo
Han, Dongyoon
author_facet Gu, Geonmo
Heo, Byeongho
Yu, Jaemyung
Hwang, Jaehui
Kim, Taekyung
Lee, Sangmin
Jun, HeeJae
Kang, Yoohoon
Yun, Sangdoo
Han, Dongyoon
contents Universal Multimodal embedding models built on Multimodal Large Language Models (MLLMs) have traditionally employed contrastive learning, which aligns representations of query-target pairs across different modalities. Yet, despite its empirical success, they are primarily built on a "single-turn" formulation where each query-target pair is treated as an independent data point. This paradigm leads to computational inefficiency when scaling, as it requires a separate forward pass for each pair and overlooks potential contextual relationships between multiple queries that can relate to the same context. In this work, we introduce Multi-Turn Contrastive Learning (MuCo), a dialogue-inspired framework that revisits this process. MuCo leverages the conversational nature of MLLMs to process multiple, related query-target pairs associated with a single image within a single forward pass. This allows us to extract a set of multiple query and target embeddings simultaneously, conditioned on a shared context representation, amplifying the effective batch size and overall training efficiency. Experiments exhibit MuCo with a newly curated 5M multimodal multi-turn dataset (M3T), which yields state-of-the-art retrieval performance on MMEB and M-BEIR benchmarks, while markedly enhancing both training efficiency and representation coherence across modalities. Code and M3T are available at https://github.com/naver-ai/muco
format Preprint
id arxiv_https___arxiv_org_abs_2602_06393
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MuCo: Multi-turn Contrastive Learning for Multimodal Embedding Model
Gu, Geonmo
Heo, Byeongho
Yu, Jaemyung
Hwang, Jaehui
Kim, Taekyung
Lee, Sangmin
Jun, HeeJae
Kang, Yoohoon
Yun, Sangdoo
Han, Dongyoon
Information Retrieval
Universal Multimodal embedding models built on Multimodal Large Language Models (MLLMs) have traditionally employed contrastive learning, which aligns representations of query-target pairs across different modalities. Yet, despite its empirical success, they are primarily built on a "single-turn" formulation where each query-target pair is treated as an independent data point. This paradigm leads to computational inefficiency when scaling, as it requires a separate forward pass for each pair and overlooks potential contextual relationships between multiple queries that can relate to the same context. In this work, we introduce Multi-Turn Contrastive Learning (MuCo), a dialogue-inspired framework that revisits this process. MuCo leverages the conversational nature of MLLMs to process multiple, related query-target pairs associated with a single image within a single forward pass. This allows us to extract a set of multiple query and target embeddings simultaneously, conditioned on a shared context representation, amplifying the effective batch size and overall training efficiency. Experiments exhibit MuCo with a newly curated 5M multimodal multi-turn dataset (M3T), which yields state-of-the-art retrieval performance on MMEB and M-BEIR benchmarks, while markedly enhancing both training efficiency and representation coherence across modalities. Code and M3T are available at https://github.com/naver-ai/muco
title MuCo: Multi-turn Contrastive Learning for Multimodal Embedding Model
topic Information Retrieval
url https://arxiv.org/abs/2602.06393