@inproceedings{zhang-etal-2026-survey-training,
title = "A Survey of the Training Process of {LLM}-Based Agents",
author = "Zhang, Zhisong and
Liu, Lemao and
Fang, Tianqing and
Cai, Deng and
Xu, Jiahao",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the {A}sia-Pacific Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.aacl-long.35/",
pages = "719--741",
ISBN = "979-8-89176-439-2",
abstract = "Autonomous agents based on large language models (LLMs) are becoming an increasingly prevalent paradigm for tackling complex and real-world tasks. Despite the remarkable zero-shot capabilities of modern LLMs, specialized agent training is often essential to obtain reliable and improved performance for specific target tasks. In this survey, we provide a process-centric review of the training of LLM-based agents, with a primary focus on methods that update the parameters of the main agent policy. Rather than organizing prior work only by application domain, agent capability, or learning algorithm, we examine the end-to-end pipeline through which LLM-based agents are trained: environment setup, data and trajectory construction, learning-signal design, and optimization objectives and schemes. This perspective highlights how design choices at different stages interact with one another. Finally, we conclude with a discussion on potential future directions."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhang-etal-2026-survey-training">
<titleInfo>
<title>A Survey of the Training Process of LLM-Based Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhisong</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lemao</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tianqing</namePart>
<namePart type="family">Fang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Deng</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiahao</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-439-2</identifier>
</relatedItem>
<abstract>Autonomous agents based on large language models (LLMs) are becoming an increasingly prevalent paradigm for tackling complex and real-world tasks. Despite the remarkable zero-shot capabilities of modern LLMs, specialized agent training is often essential to obtain reliable and improved performance for specific target tasks. In this survey, we provide a process-centric review of the training of LLM-based agents, with a primary focus on methods that update the parameters of the main agent policy. Rather than organizing prior work only by application domain, agent capability, or learning algorithm, we examine the end-to-end pipeline through which LLM-based agents are trained: environment setup, data and trajectory construction, learning-signal design, and optimization objectives and schemes. This perspective highlights how design choices at different stages interact with one another. Finally, we conclude with a discussion on potential future directions.</abstract>
<identifier type="citekey">zhang-etal-2026-survey-training</identifier>
<location>
<url>https://aclanthology.org/2026.aacl-long.35/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>719</start>
<end>741</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T A Survey of the Training Process of LLM-Based Agents
%A Zhang, Zhisong
%A Liu, Lemao
%A Fang, Tianqing
%A Cai, Deng
%A Xu, Jiahao
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-439-2
%F zhang-etal-2026-survey-training
%X Autonomous agents based on large language models (LLMs) are becoming an increasingly prevalent paradigm for tackling complex and real-world tasks. Despite the remarkable zero-shot capabilities of modern LLMs, specialized agent training is often essential to obtain reliable and improved performance for specific target tasks. In this survey, we provide a process-centric review of the training of LLM-based agents, with a primary focus on methods that update the parameters of the main agent policy. Rather than organizing prior work only by application domain, agent capability, or learning algorithm, we examine the end-to-end pipeline through which LLM-based agents are trained: environment setup, data and trajectory construction, learning-signal design, and optimization objectives and schemes. This perspective highlights how design choices at different stages interact with one another. Finally, we conclude with a discussion on potential future directions.
%U https://aclanthology.org/2026.aacl-long.35/
%P 719-741
Markdown (Informal)
[A Survey of the Training Process of LLM-Based Agents](https://aclanthology.org/2026.aacl-long.35/) (Zhang et al., AACL-IJCNLP 2026)
ACL
- Zhisong Zhang, Lemao Liu, Tianqing Fang, Deng Cai, and Jiahao Xu. 2026. A Survey of the Training Process of LLM-Based Agents. In Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 719–741, Hengqin, China. Association for Computational Linguistics.