<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="carvalho-etal-2025-langvae">
    <titleInfo>
        <title>LangVAE and LangSpace: Building and Probing for Language Model VAEs</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Danilo</namePart>
        <namePart type="family">Carvalho</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yingji</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Harriet</namePart>
        <namePart type="family">Unsworth</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Andre</namePart>
        <namePart type="family">Freitas</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: System Demonstrations</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Ivan</namePart>
            <namePart type="family">Habernal</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Peter</namePart>
            <namePart type="family">Schulam</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jörg</namePart>
            <namePart type="family">Tiedemann</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">979-8-89176-334-0</identifier>
    </relatedItem>
    <abstract>We present LangVAE, a novel framework for modular construction of variational autoencoders (VAEs) on top of pre-trained large language models (LLMs). Such language model VAEs can encode the knowledge of their pre-trained components into more compact and semantically disentangled representations. The representations obtained in this way can be analysed with the LangVAE companion framework: LangSpace, which implements a collection of probing methods, such as vector traversal and interpolation, disentanglement measures, and cluster visualisations. LangVAE and LangSpace offer a flexible, efficient and scalable way of building and analysing textual representations, with simple integration for models available on the HuggingFace Hub. Additionally, we conducted a set of experiments with different encoder and decoder combinations, as well as annotated inputs, revealing a wide range of interactions across architectural families and sizes w.r.t.generalisation and disentanglement. Our findings demonstrate a promising framework for systematising the experimentation and understanding of textual representations.</abstract>
    <identifier type="citekey">carvalho-etal-2025-langvae</identifier>
    <location>
        <url>https://aclanthology.org/2025.emnlp-demos.57/</url>
    </location>
    <part>
        <date>2025-11</date>
        <extent unit="page">
            <start>749</start>
            <end>759</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T LangVAE and LangSpace: Building and Probing for Language Model VAEs
%A Carvalho, Danilo
%A Zhang, Yingji
%A Unsworth, Harriet
%A Freitas, Andre
%Y Habernal, Ivan
%Y Schulam, Peter
%Y Tiedemann, Jörg
%S Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-334-0
%F carvalho-etal-2025-langvae
%X We present LangVAE, a novel framework for modular construction of variational autoencoders (VAEs) on top of pre-trained large language models (LLMs). Such language model VAEs can encode the knowledge of their pre-trained components into more compact and semantically disentangled representations. The representations obtained in this way can be analysed with the LangVAE companion framework: LangSpace, which implements a collection of probing methods, such as vector traversal and interpolation, disentanglement measures, and cluster visualisations. LangVAE and LangSpace offer a flexible, efficient and scalable way of building and analysing textual representations, with simple integration for models available on the HuggingFace Hub. Additionally, we conducted a set of experiments with different encoder and decoder combinations, as well as annotated inputs, revealing a wide range of interactions across architectural families and sizes w.r.t.generalisation and disentanglement. Our findings demonstrate a promising framework for systematising the experimentation and understanding of textual representations.
%U https://aclanthology.org/2025.emnlp-demos.57/
%P 749-759

Markdown (Informal)

[LangVAE and LangSpace: Building and Probing for Language Model VAEs](https://aclanthology.org/2025.emnlp-demos.57/) (Carvalho et al., EMNLP 2025)

• LangVAE and LangSpace: Building and Probing for Language Model VAEs (Carvalho et al., EMNLP 2025)

ACL

• Danilo Carvalho, Yingji Zhang, Harriet Unsworth, and Andre Freitas. 2025. LangVAE and LangSpace: Building and Probing for Language Model VAEs. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, pages 749–759, Suzhou, China. Association for Computational Linguistics.