@inproceedings{wu-etal-2026-building,
title = "Building {LLM}s Like {LEGO}: Two-dimensional Architecture Reassembly of Large Language Models",
author = "Wu, Xingyu and
Zhou, Yu and
Tan, KC",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.acl-long.2081/",
pages = "44939--44956",
ISBN = "979-8-89176-390-6",
abstract = "Pretrained large language models (LLMs) are typically reused as indivisible artifacts, adapted, merged, or ensembled as a whole. In this study, we show that LLMs can instead be structurally recomposed as modular building blocks to create new architectures without access to original training data. We introduce architecture-level reassembly as a new reuse paradigm, in which Transformer blocks from heterogeneous models are treated as reusable components. This idea is formalized through a two-dimensional reassembly space that supports both vertical recombination across depth and horizontal composition within layers. To make this space tractable, we propose a chromosome-based architectural encoding and perform a bi-level multi-objective evolutionary optimization over vertical structure and horizontal composition. To resolve representation incompatibility across heterogeneous blocks, we introduce lightweight glue layers trained via data-free knowledge distillation, enabling valid information flow without modifying pretrained parameters. Our results demonstrate that architecture-level reassembly unlocks a new dimension of flexibility in model reuse, pointing toward a modular and evolutionary view of LLM design."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2026-building">
<titleInfo>
<title>Building LLMs Like LEGO: Two-dimensional Architecture Reassembly of Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Xingyu</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">K</namePart>
<namePart type="given">C</namePart>
<namePart type="family">Tan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-390-6</identifier>
</relatedItem>
<abstract>Pretrained large language models (LLMs) are typically reused as indivisible artifacts, adapted, merged, or ensembled as a whole. In this study, we show that LLMs can instead be structurally recomposed as modular building blocks to create new architectures without access to original training data. We introduce architecture-level reassembly as a new reuse paradigm, in which Transformer blocks from heterogeneous models are treated as reusable components. This idea is formalized through a two-dimensional reassembly space that supports both vertical recombination across depth and horizontal composition within layers. To make this space tractable, we propose a chromosome-based architectural encoding and perform a bi-level multi-objective evolutionary optimization over vertical structure and horizontal composition. To resolve representation incompatibility across heterogeneous blocks, we introduce lightweight glue layers trained via data-free knowledge distillation, enabling valid information flow without modifying pretrained parameters. Our results demonstrate that architecture-level reassembly unlocks a new dimension of flexibility in model reuse, pointing toward a modular and evolutionary view of LLM design.</abstract>
<identifier type="citekey">wu-etal-2026-building</identifier>
<location>
<url>https://aclanthology.org/2026.acl-long.2081/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>44939</start>
<end>44956</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Building LLMs Like LEGO: Two-dimensional Architecture Reassembly of Large Language Models
%A Wu, Xingyu
%A Zhou, Yu
%A Tan, K. C.
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-390-6
%F wu-etal-2026-building
%X Pretrained large language models (LLMs) are typically reused as indivisible artifacts, adapted, merged, or ensembled as a whole. In this study, we show that LLMs can instead be structurally recomposed as modular building blocks to create new architectures without access to original training data. We introduce architecture-level reassembly as a new reuse paradigm, in which Transformer blocks from heterogeneous models are treated as reusable components. This idea is formalized through a two-dimensional reassembly space that supports both vertical recombination across depth and horizontal composition within layers. To make this space tractable, we propose a chromosome-based architectural encoding and perform a bi-level multi-objective evolutionary optimization over vertical structure and horizontal composition. To resolve representation incompatibility across heterogeneous blocks, we introduce lightweight glue layers trained via data-free knowledge distillation, enabling valid information flow without modifying pretrained parameters. Our results demonstrate that architecture-level reassembly unlocks a new dimension of flexibility in model reuse, pointing toward a modular and evolutionary view of LLM design.
%U https://aclanthology.org/2026.acl-long.2081/
%P 44939-44956
Markdown (Informal)
[Building LLMs Like LEGO: Two-dimensional Architecture Reassembly of Large Language Models](https://aclanthology.org/2026.acl-long.2081/) (Wu et al., ACL 2026)
ACL