<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="geigle-etal-2024-mblip">
    <titleInfo>
        <title>mBLIP: Efficient Bootstrapping of Multilingual Vision-LLMs</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Gregor</namePart>
        <namePart type="family">Geigle</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Abhay</namePart>
        <namePart type="family">Jain</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Radu</namePart>
        <namePart type="family">Timofte</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Goran</namePart>
        <namePart type="family">Glavaš</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2024-08</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 3rd Workshop on Advances in Language and Vision Research (ALVR)</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Jing</namePart>
            <namePart type="family">Gu</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Tsu-Jui</namePart>
            <namePart type="given">(Ray)</namePart>
            <namePart type="family">Fu</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Drew</namePart>
            <namePart type="family">Hudson</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Asli</namePart>
            <namePart type="family">Celikyilmaz</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">William</namePart>
            <namePart type="family">Wang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Bangkok, Thailand</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Modular vision-language models (Vision-LLMs) align pretrained image encoders with (frozen) large language models (LLMs) and post-hoc condition LLMs to ‘understand’ the image input. With the abundance of readily available high-quality English image-text data as well as strong monolingual English LLMs, the research focus has been on English-only Vision-LLMs. Multilingual vision-language models are still predominantly obtained via expensive end-to-end pretraining, resulting in comparatively smaller models, trained on limited multilingual image data supplemented with text-only multilingual corpora. We present mBLIP, the first Vision-LLM leveraging multilingual LLMs, which we obtain in a computationally efficient manner on consumer-level hardware. To this end, we re-align an image encoder previously tuned to an English LLM to a new, multilingual LLM using only a few million multilingual training examples derived from a mix of vision-and-language tasks, which we obtain by machine-translating high-quality English data to 95 languages. On the IGLUE benchmark and XM3600, mBLIP yields results competitive with state-of-the-art models and it greatly outperforms strong English-only Vision-LLMs like Llava 1.5. We release our model, code, and train data at https://github.com/gregor-ge/mBLIP.</abstract>
    <identifier type="citekey">geigle-etal-2024-mblip</identifier>
    <identifier type="doi">10.18653/v1/2024.alvr-1.2</identifier>
    <location>
        <url>https://aclanthology.org/2024.alvr-1.2/</url>
    </location>
    <part>
        <date>2024-08</date>
        <extent unit="page">
            <start>7</start>
            <end>25</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T mBLIP: Efficient Bootstrapping of Multilingual Vision-LLMs
%A Geigle, Gregor
%A Jain, Abhay
%A Timofte, Radu
%A Glavaš, Goran
%Y Gu, Jing
%Y Fu, Tsu-Jui (Ray)
%Y Hudson, Drew
%Y Celikyilmaz, Asli
%Y Wang, William
%S Proceedings of the 3rd Workshop on Advances in Language and Vision Research (ALVR)
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F geigle-etal-2024-mblip
%X Modular vision-language models (Vision-LLMs) align pretrained image encoders with (frozen) large language models (LLMs) and post-hoc condition LLMs to ‘understand’ the image input. With the abundance of readily available high-quality English image-text data as well as strong monolingual English LLMs, the research focus has been on English-only Vision-LLMs. Multilingual vision-language models are still predominantly obtained via expensive end-to-end pretraining, resulting in comparatively smaller models, trained on limited multilingual image data supplemented with text-only multilingual corpora. We present mBLIP, the first Vision-LLM leveraging multilingual LLMs, which we obtain in a computationally efficient manner on consumer-level hardware. To this end, we re-align an image encoder previously tuned to an English LLM to a new, multilingual LLM using only a few million multilingual training examples derived from a mix of vision-and-language tasks, which we obtain by machine-translating high-quality English data to 95 languages. On the IGLUE benchmark and XM3600, mBLIP yields results competitive with state-of-the-art models and it greatly outperforms strong English-only Vision-LLMs like Llava 1.5. We release our model, code, and train data at https://github.com/gregor-ge/mBLIP.
%R 10.18653/v1/2024.alvr-1.2
%U https://aclanthology.org/2024.alvr-1.2/
%U https://doi.org/10.18653/v1/2024.alvr-1.2
%P 7-25

Markdown (Informal)

[mBLIP: Efficient Bootstrapping of Multilingual Vision-LLMs](https://aclanthology.org/2024.alvr-1.2/) (Geigle et al., ALVR 2024)

• mBLIP: Efficient Bootstrapping of Multilingual Vision-LLMs (Geigle et al., ALVR 2024)

ACL

• Gregor Geigle, Abhay Jain, Radu Timofte, and Goran Glavaš. 2024. mBLIP: Efficient Bootstrapping of Multilingual Vision-LLMs. In Proceedings of the 3rd Workshop on Advances in Language and Vision Research (ALVR), pages 7–25, Bangkok, Thailand. Association for Computational Linguistics.