@inproceedings{lee-etal-2026-grad,
title = "{GRAD}: Generalizing {RAG} Adaptation with Decoding",
author = "Lee, Youngwon and
Hwang, Seung-won and
Yao, Zhewei and
He, Yuxiong",
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.2099/",
doi = "10.18653/v1/2026.acl-long.2099",
pages = "45274--45290",
ISBN = "979-8-89176-390-6",
abstract = "Retrieval-augmented generation needs generation to follow retrieved evidence across shifting domains and prompt layouts, but training a new stronger model per task is costly. To this end, we propose GRAD, an adaptive decoding-time framework that keeps the base generator fixed and composes small, objective-specific guidance at inference. A key advantage of this design is enabling mix and match diverse RAG objectives: model scaling (MS), domain adaptation (DA) and positional debiasing (DB) can be integrated as token-level guidance terms, and new objectives can be easily plugged in. Across public benchmarks and private settings with no in-domain labels, GRAD improves accuracy with favorable latency, offering strong trade-offs versus scaling while reliably activating helpful objectives and suppressing harmful ones, adaptively to tasks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lee-etal-2026-grad">
<titleInfo>
<title>GRAD: Generalizing RAG Adaptation with Decoding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Youngwon</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seung-won</namePart>
<namePart type="family">Hwang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhewei</namePart>
<namePart type="family">Yao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuxiong</namePart>
<namePart type="family">He</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>Retrieval-augmented generation needs generation to follow retrieved evidence across shifting domains and prompt layouts, but training a new stronger model per task is costly. To this end, we propose GRAD, an adaptive decoding-time framework that keeps the base generator fixed and composes small, objective-specific guidance at inference. A key advantage of this design is enabling mix and match diverse RAG objectives: model scaling (MS), domain adaptation (DA) and positional debiasing (DB) can be integrated as token-level guidance terms, and new objectives can be easily plugged in. Across public benchmarks and private settings with no in-domain labels, GRAD improves accuracy with favorable latency, offering strong trade-offs versus scaling while reliably activating helpful objectives and suppressing harmful ones, adaptively to tasks.</abstract>
<identifier type="citekey">lee-etal-2026-grad</identifier>
<identifier type="doi">10.18653/v1/2026.acl-long.2099</identifier>
<location>
<url>https://aclanthology.org/2026.acl-long.2099/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>45274</start>
<end>45290</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T GRAD: Generalizing RAG Adaptation with Decoding
%A Lee, Youngwon
%A Hwang, Seung-won
%A Yao, Zhewei
%A He, Yuxiong
%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 lee-etal-2026-grad
%X Retrieval-augmented generation needs generation to follow retrieved evidence across shifting domains and prompt layouts, but training a new stronger model per task is costly. To this end, we propose GRAD, an adaptive decoding-time framework that keeps the base generator fixed and composes small, objective-specific guidance at inference. A key advantage of this design is enabling mix and match diverse RAG objectives: model scaling (MS), domain adaptation (DA) and positional debiasing (DB) can be integrated as token-level guidance terms, and new objectives can be easily plugged in. Across public benchmarks and private settings with no in-domain labels, GRAD improves accuracy with favorable latency, offering strong trade-offs versus scaling while reliably activating helpful objectives and suppressing harmful ones, adaptively to tasks.
%R 10.18653/v1/2026.acl-long.2099
%U https://aclanthology.org/2026.acl-long.2099/
%U https://doi.org/10.18653/v1/2026.acl-long.2099
%P 45274-45290
Markdown (Informal)
[GRAD: Generalizing RAG Adaptation with Decoding](https://aclanthology.org/2026.acl-long.2099/) (Lee et al., ACL 2026)
ACL
- Youngwon Lee, Seung-won Hwang, Zhewei Yao, and Yuxiong He. 2026. GRAD: Generalizing RAG Adaptation with Decoding. In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 45274–45290, San Diego, California, United States. Association for Computational Linguistics.