@inproceedings{shirai-etal-2026-generalization,
title = "Generalization in Graph Reasoning: A Systematic Comparison of {LLM} Training Approaches",
author = "Shirai, Sola and
Srinivas, Kavitha and
Dolby, Julian and
Katz, Michael and
Sohrabi, Shirin and
Samulowitz, Horst",
editor = "Gupta, Vivek and
Ding, Kaize and
Kokel, Harsha and
Zhao, Yue and
Agarwal, Amit and
Wang, Yu and
Glass, Michael and
Zhang, Yu and
Srinivas, Kavitha and
Chen, Xiusi and
Hassanzadeh, Oktie and
Zhu, Qi and
Chang, Shuaichen and
Luo, Yuan",
booktitle = "Proceedings of the First Workshop on Structured Understanding, Retrieval, and Generation in the {LLM} Era ({SURG}e{LLM} 2026)",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.surgellm-1.18/",
pages = "275--297",
ISBN = "979-8-89176-406-4",
abstract = "For large language models (LLMs), reasoning over graphs can help solve many problems. Prior work has tried to improve LLM graph reasoning through different training methods, but the merits of such approaches remain unclear and the limitations of each approach with respect to generalizability of reasoning are often not thoroughly explored. In this paper we systematically compare the ability of LLMs to learn fundamental graph tasks across a variety of training methods and their ability to generalize out of distribution across various dimensions. We highlight key tradeoffs between training methods, e.g., training specialized graph encoders and fusing their embeddings with LLMs consistently collapses in terms of generalizability; however, no single method shows clear superiority across all dimensions of generalizability, regardless of the size of the model."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shirai-etal-2026-generalization">
<titleInfo>
<title>Generalization in Graph Reasoning: A Systematic Comparison of LLM Training Approaches</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sola</namePart>
<namePart type="family">Shirai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kavitha</namePart>
<namePart type="family">Srinivas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Julian</namePart>
<namePart type="family">Dolby</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michael</namePart>
<namePart type="family">Katz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shirin</namePart>
<namePart type="family">Sohrabi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Horst</namePart>
<namePart type="family">Samulowitz</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 First Workshop on Structured Understanding, Retrieval, and Generation in the LLM Era (SURGeLLM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kaize</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Harsha</namePart>
<namePart type="family">Kokel</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Amit</namePart>
<namePart type="family">Agarwal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michael</namePart>
<namePart type="family">Glass</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kavitha</namePart>
<namePart type="family">Srinivas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiusi</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Oktie</namePart>
<namePart type="family">Hassanzadeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qi</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shuaichen</namePart>
<namePart type="family">Chang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuan</namePart>
<namePart type="family">Luo</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">979-8-89176-406-4</identifier>
</relatedItem>
<abstract>For large language models (LLMs), reasoning over graphs can help solve many problems. Prior work has tried to improve LLM graph reasoning through different training methods, but the merits of such approaches remain unclear and the limitations of each approach with respect to generalizability of reasoning are often not thoroughly explored. In this paper we systematically compare the ability of LLMs to learn fundamental graph tasks across a variety of training methods and their ability to generalize out of distribution across various dimensions. We highlight key tradeoffs between training methods, e.g., training specialized graph encoders and fusing their embeddings with LLMs consistently collapses in terms of generalizability; however, no single method shows clear superiority across all dimensions of generalizability, regardless of the size of the model.</abstract>
<identifier type="citekey">shirai-etal-2026-generalization</identifier>
<location>
<url>https://aclanthology.org/2026.surgellm-1.18/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>275</start>
<end>297</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Generalization in Graph Reasoning: A Systematic Comparison of LLM Training Approaches
%A Shirai, Sola
%A Srinivas, Kavitha
%A Dolby, Julian
%A Katz, Michael
%A Sohrabi, Shirin
%A Samulowitz, Horst
%Y Gupta, Vivek
%Y Ding, Kaize
%Y Kokel, Harsha
%Y Zhao, Yue
%Y Agarwal, Amit
%Y Wang, Yu
%Y Glass, Michael
%Y Zhang, Yu
%Y Srinivas, Kavitha
%Y Chen, Xiusi
%Y Hassanzadeh, Oktie
%Y Zhu, Qi
%Y Chang, Shuaichen
%Y Luo, Yuan
%S Proceedings of the First Workshop on Structured Understanding, Retrieval, and Generation in the LLM Era (SURGeLLM 2026)
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-406-4
%F shirai-etal-2026-generalization
%X For large language models (LLMs), reasoning over graphs can help solve many problems. Prior work has tried to improve LLM graph reasoning through different training methods, but the merits of such approaches remain unclear and the limitations of each approach with respect to generalizability of reasoning are often not thoroughly explored. In this paper we systematically compare the ability of LLMs to learn fundamental graph tasks across a variety of training methods and their ability to generalize out of distribution across various dimensions. We highlight key tradeoffs between training methods, e.g., training specialized graph encoders and fusing their embeddings with LLMs consistently collapses in terms of generalizability; however, no single method shows clear superiority across all dimensions of generalizability, regardless of the size of the model.
%U https://aclanthology.org/2026.surgellm-1.18/
%P 275-297
Markdown (Informal)
[Generalization in Graph Reasoning: A Systematic Comparison of LLM Training Approaches](https://aclanthology.org/2026.surgellm-1.18/) (Shirai et al., SURGeLLM 2026)
ACL
- Sola Shirai, Kavitha Srinivas, Julian Dolby, Michael Katz, Shirin Sohrabi, and Horst Samulowitz. 2026. Generalization in Graph Reasoning: A Systematic Comparison of LLM Training Approaches. In Proceedings of the First Workshop on Structured Understanding, Retrieval, and Generation in the LLM Era (SURGeLLM 2026), pages 275–297, San Diego, California, United States. Association for Computational Linguistics.