@inproceedings{koul-etal-2026-robust,
title = "Robust In-Context Selection via Online Learned Position-Corrected Attention",
author = "Koul, Deeksha and
Kumar, Gaurav and
Sabale, Yash and
Sarawagi, Sunita",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1743/",
doi = "10.18653/v1/2026.findings-acl.1743",
pages = "34919--34930",
ISBN = "979-8-89176-395-1",
abstract = "Large Language Models (LLMs) are often deployed in tasks that require selecting an item from a long list provided in the model{'}s context. LLMs' native selection behavior is brittle: predictions are sensitive to the surface form of the identifiers, their placement within the context, and the ordering of candidate items. We present OLR-Heads, a robust method for list selection that harnesses attention patterns available from a single forward call on the LLM. OLR-Heads learns the logic for item selection using a few in-context examples, and a simple online position-debiasing mechanism to correct attention distortion. Across multiple database and tool selection benchmarks, OLR-Heads consistently improves selection performance over direct generation and prior attention-based methods, while remaining robust to prompt variations and item ordering.The LLM{'}s KV cache states are unaffected, and can be reused for subsequent response generation. In contrast, existing approaches either entail additional LLM calls, or task-specific offline learning, or position debiasing methods that modify the attention or encoding rendering the KV states unusable for subsequent generation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="koul-etal-2026-robust">
<titleInfo>
<title>Robust In-Context Selection via Online Learned Position-Corrected Attention</title>
</titleInfo>
<name type="personal">
<namePart type="given">Deeksha</namePart>
<namePart type="family">Koul</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gaurav</namePart>
<namePart type="family">Kumar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yash</namePart>
<namePart type="family">Sabale</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunita</namePart>
<namePart type="family">Sarawagi</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>Findings of the Association for Computational Linguistics: ACL 2026</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">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) are often deployed in tasks that require selecting an item from a long list provided in the model’s context. LLMs’ native selection behavior is brittle: predictions are sensitive to the surface form of the identifiers, their placement within the context, and the ordering of candidate items. We present OLR-Heads, a robust method for list selection that harnesses attention patterns available from a single forward call on the LLM. OLR-Heads learns the logic for item selection using a few in-context examples, and a simple online position-debiasing mechanism to correct attention distortion. Across multiple database and tool selection benchmarks, OLR-Heads consistently improves selection performance over direct generation and prior attention-based methods, while remaining robust to prompt variations and item ordering.The LLM’s KV cache states are unaffected, and can be reused for subsequent response generation. In contrast, existing approaches either entail additional LLM calls, or task-specific offline learning, or position debiasing methods that modify the attention or encoding rendering the KV states unusable for subsequent generation.</abstract>
<identifier type="citekey">koul-etal-2026-robust</identifier>
<identifier type="doi">10.18653/v1/2026.findings-acl.1743</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1743/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>34919</start>
<end>34930</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Robust In-Context Selection via Online Learned Position-Corrected Attention
%A Koul, Deeksha
%A Kumar, Gaurav
%A Sabale, Yash
%A Sarawagi, Sunita
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F koul-etal-2026-robust
%X Large Language Models (LLMs) are often deployed in tasks that require selecting an item from a long list provided in the model’s context. LLMs’ native selection behavior is brittle: predictions are sensitive to the surface form of the identifiers, their placement within the context, and the ordering of candidate items. We present OLR-Heads, a robust method for list selection that harnesses attention patterns available from a single forward call on the LLM. OLR-Heads learns the logic for item selection using a few in-context examples, and a simple online position-debiasing mechanism to correct attention distortion. Across multiple database and tool selection benchmarks, OLR-Heads consistently improves selection performance over direct generation and prior attention-based methods, while remaining robust to prompt variations and item ordering.The LLM’s KV cache states are unaffected, and can be reused for subsequent response generation. In contrast, existing approaches either entail additional LLM calls, or task-specific offline learning, or position debiasing methods that modify the attention or encoding rendering the KV states unusable for subsequent generation.
%R 10.18653/v1/2026.findings-acl.1743
%U https://aclanthology.org/2026.findings-acl.1743/
%U https://doi.org/10.18653/v1/2026.findings-acl.1743
%P 34919-34930
Markdown (Informal)
[Robust In-Context Selection via Online Learned Position-Corrected Attention](https://aclanthology.org/2026.findings-acl.1743/) (Koul et al., Findings 2026)
ACL