@inproceedings{maraj-sethi-2026-query,
title = "Do Query-Agnostic Memory Structures Earn Their Budget? A 10{M}-Token Audit with Exact Read Limits",
author = "Maraj, Amit and
Sethi, Kannav",
editor = "Kamalloo, Ehsan and
Gontier, Nicolas and
Rozanov, Nikolai and
Hashemzadeh, Maryam and
Arora, Akhil",
booktitle = "Proceedings of the Second Workshop for {RE}search on Agent Language Models ({REALM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.realm-1.18/",
pages = "263--270",
ISBN = "979-8-89176-485-9",
abstract = "Long-term memory systems often organize experience before future queries are known, but whether this improves evidence access under a fixed read budget remains unclear. We evaluate inexpensive query-agnostic segmentation and hierarchical retrieval on AMA-Bench agent trajectories and BEAM conversations spanning 128K to 10M tokens. Experiments use exact serialized-context budgets, development-only tuning, two sentence encoders, and uncertainty estimates clustered by trajectory or conversation. In an exploratory post-plan analysis of 1,117 step-anchored trajectory questions, fixed three-turn windows outperform all tested alternative boundary methods. On BEAM, primary-encoder flat retrieval recall falls from 0.564 at 128K to 0.331 at 10M. Single-vector hierarchical routing further reduces recall; multiple local vectors recover part of this loss but do not surpass flat retrieval. Doubling the read budget from 8,000 to 16,000 characters improves evidence recall at every scale with both encoders. However, the 10M evaluation, comprising 130 questions from eight conversations, detects no corresponding QA improvement. Single-vector routing reduces QA accuracy, while multi-vector routing is not distinguishable from flat retrieval. These findings identify coarse representation quality as a bottleneck in the tested hierarchies and show that improved evidence access need not yield better answers."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="maraj-sethi-2026-query">
<titleInfo>
<title>Do Query-Agnostic Memory Structures Earn Their Budget? A 10M-Token Audit with Exact Read Limits</title>
</titleInfo>
<name type="personal">
<namePart type="given">Amit</namePart>
<namePart type="family">Maraj</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kannav</namePart>
<namePart type="family">Sethi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ehsan</namePart>
<namePart type="family">Kamalloo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicolas</namePart>
<namePart type="family">Gontier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikolai</namePart>
<namePart type="family">Rozanov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maryam</namePart>
<namePart type="family">Hashemzadeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Akhil</namePart>
<namePart type="family">Arora</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-485-9</identifier>
</relatedItem>
<abstract>Long-term memory systems often organize experience before future queries are known, but whether this improves evidence access under a fixed read budget remains unclear. We evaluate inexpensive query-agnostic segmentation and hierarchical retrieval on AMA-Bench agent trajectories and BEAM conversations spanning 128K to 10M tokens. Experiments use exact serialized-context budgets, development-only tuning, two sentence encoders, and uncertainty estimates clustered by trajectory or conversation. In an exploratory post-plan analysis of 1,117 step-anchored trajectory questions, fixed three-turn windows outperform all tested alternative boundary methods. On BEAM, primary-encoder flat retrieval recall falls from 0.564 at 128K to 0.331 at 10M. Single-vector hierarchical routing further reduces recall; multiple local vectors recover part of this loss but do not surpass flat retrieval. Doubling the read budget from 8,000 to 16,000 characters improves evidence recall at every scale with both encoders. However, the 10M evaluation, comprising 130 questions from eight conversations, detects no corresponding QA improvement. Single-vector routing reduces QA accuracy, while multi-vector routing is not distinguishable from flat retrieval. These findings identify coarse representation quality as a bottleneck in the tested hierarchies and show that improved evidence access need not yield better answers.</abstract>
<identifier type="citekey">maraj-sethi-2026-query</identifier>
<location>
<url>https://aclanthology.org/2026.realm-1.18/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>263</start>
<end>270</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Do Query-Agnostic Memory Structures Earn Their Budget? A 10M-Token Audit with Exact Read Limits
%A Maraj, Amit
%A Sethi, Kannav
%Y Kamalloo, Ehsan
%Y Gontier, Nicolas
%Y Rozanov, Nikolai
%Y Hashemzadeh, Maryam
%Y Arora, Akhil
%S Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-485-9
%F maraj-sethi-2026-query
%X Long-term memory systems often organize experience before future queries are known, but whether this improves evidence access under a fixed read budget remains unclear. We evaluate inexpensive query-agnostic segmentation and hierarchical retrieval on AMA-Bench agent trajectories and BEAM conversations spanning 128K to 10M tokens. Experiments use exact serialized-context budgets, development-only tuning, two sentence encoders, and uncertainty estimates clustered by trajectory or conversation. In an exploratory post-plan analysis of 1,117 step-anchored trajectory questions, fixed three-turn windows outperform all tested alternative boundary methods. On BEAM, primary-encoder flat retrieval recall falls from 0.564 at 128K to 0.331 at 10M. Single-vector hierarchical routing further reduces recall; multiple local vectors recover part of this loss but do not surpass flat retrieval. Doubling the read budget from 8,000 to 16,000 characters improves evidence recall at every scale with both encoders. However, the 10M evaluation, comprising 130 questions from eight conversations, detects no corresponding QA improvement. Single-vector routing reduces QA accuracy, while multi-vector routing is not distinguishable from flat retrieval. These findings identify coarse representation quality as a bottleneck in the tested hierarchies and show that improved evidence access need not yield better answers.
%U https://aclanthology.org/2026.realm-1.18/
%P 263-270
Markdown (Informal)
[Do Query-Agnostic Memory Structures Earn Their Budget? A 10M-Token Audit with Exact Read Limits](https://aclanthology.org/2026.realm-1.18/) (Maraj & Sethi, REALM 2026)
ACL