@inproceedings{suzgun-etal-2026-dynamic,
title = "Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory",
author = "Suzgun, Mirac and
Yuksekgonul, Mert and
Bianchi, Federico and
Jurafsky, Dan and
Zou, James",
editor = "Demberg, Vera and
Inui, Kentaro and
Marquez, Llu{\'i}s",
booktitle = "Proceedings of the 19th Conference of the {E}uropean Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = mar,
year = "2026",
address = "Rabat, Morocco",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.eacl-long.333/",
pages = "7080--7106",
ISBN = "979-8-89176-380-7",
abstract = "Despite their impressive performance on complex tasks, current language models (LMs) typically operate in a vacuum: Each input query is processed separately, without retaining insights from previous attempts. Here, we present Dynamic Cheatsheet (DC), a lightweight framework that endows a black-box LM with a persistent, evolving memory. Rather than repeatedly re-discovering or re-committing the same solutions and mistakes, DC enables models to store and reuse accumulated strategies, code snippets, and general problem-solving insights at inference time. This test-time learning enhances performance substantially across a range of tasks without needing explicit ground-truth labels or human feedback. Leveraging DC, Claude 3.5 Sonnet{'}s accuracy more than doubled on AIME math exams once it began retaining algebraic insights across questions. Similarly, GPT-4o{'}s success rate on the Game of 24 puzzle increased from about 10{\%} to 99{\%} after the model discovered and reused a Python-based solution. In tasks prone to arithmetic mistakes, such as balancing equations, DC enabled GPT-4o and Claude to reach near-perfect accuracy by recalling previously validated code, whereas their baselines stagnated around 50{\%}. Beyond arithmetic challenges, DC yields notable accuracy gains on knowledge-demanding tasks. Claude achieved a 9{\%} improvement in GPQA-Diamond and an 8{\%} boost on MMLU-Pro Engineering and Physics problems. Crucially, DC{'}s memory is self-curated, focusing on concise, transferable snippets rather than entire transcripts, thereby facilitating meta-learning and avoiding context ballooning. Unlike fine-tuning or static retrieval methods, DC adapts LMs' problem-solving skills on the fly, without modifying their underlying parameters, and offers a practical approach for continuously refining responses and cutting routine errors. Overall, our findings present DC as a promising approach for augmenting LMs with persistent memory, bridging the divide between isolated inference events and the cumulative, experience-driven learning characteristic of human cognition."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="suzgun-etal-2026-dynamic">
<titleInfo>
<title>Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mirac</namePart>
<namePart type="family">Suzgun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mert</namePart>
<namePart type="family">Yuksekgonul</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Federico</namePart>
<namePart type="family">Bianchi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dan</namePart>
<namePart type="family">Jurafsky</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">James</namePart>
<namePart type="family">Zou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-03</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Vera</namePart>
<namePart type="family">Demberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kentaro</namePart>
<namePart type="family">Inui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lluís</namePart>
<namePart type="family">Marquez</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Rabat, Morocco</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-380-7</identifier>
</relatedItem>
<abstract>Despite their impressive performance on complex tasks, current language models (LMs) typically operate in a vacuum: Each input query is processed separately, without retaining insights from previous attempts. Here, we present Dynamic Cheatsheet (DC), a lightweight framework that endows a black-box LM with a persistent, evolving memory. Rather than repeatedly re-discovering or re-committing the same solutions and mistakes, DC enables models to store and reuse accumulated strategies, code snippets, and general problem-solving insights at inference time. This test-time learning enhances performance substantially across a range of tasks without needing explicit ground-truth labels or human feedback. Leveraging DC, Claude 3.5 Sonnet’s accuracy more than doubled on AIME math exams once it began retaining algebraic insights across questions. Similarly, GPT-4o’s success rate on the Game of 24 puzzle increased from about 10% to 99% after the model discovered and reused a Python-based solution. In tasks prone to arithmetic mistakes, such as balancing equations, DC enabled GPT-4o and Claude to reach near-perfect accuracy by recalling previously validated code, whereas their baselines stagnated around 50%. Beyond arithmetic challenges, DC yields notable accuracy gains on knowledge-demanding tasks. Claude achieved a 9% improvement in GPQA-Diamond and an 8% boost on MMLU-Pro Engineering and Physics problems. Crucially, DC’s memory is self-curated, focusing on concise, transferable snippets rather than entire transcripts, thereby facilitating meta-learning and avoiding context ballooning. Unlike fine-tuning or static retrieval methods, DC adapts LMs’ problem-solving skills on the fly, without modifying their underlying parameters, and offers a practical approach for continuously refining responses and cutting routine errors. Overall, our findings present DC as a promising approach for augmenting LMs with persistent memory, bridging the divide between isolated inference events and the cumulative, experience-driven learning characteristic of human cognition.</abstract>
<identifier type="citekey">suzgun-etal-2026-dynamic</identifier>
<location>
<url>https://aclanthology.org/2026.eacl-long.333/</url>
</location>
<part>
<date>2026-03</date>
<extent unit="page">
<start>7080</start>
<end>7106</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory
%A Suzgun, Mirac
%A Yuksekgonul, Mert
%A Bianchi, Federico
%A Jurafsky, Dan
%A Zou, James
%Y Demberg, Vera
%Y Inui, Kentaro
%Y Marquez, Lluís
%S Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 March
%I Association for Computational Linguistics
%C Rabat, Morocco
%@ 979-8-89176-380-7
%F suzgun-etal-2026-dynamic
%X Despite their impressive performance on complex tasks, current language models (LMs) typically operate in a vacuum: Each input query is processed separately, without retaining insights from previous attempts. Here, we present Dynamic Cheatsheet (DC), a lightweight framework that endows a black-box LM with a persistent, evolving memory. Rather than repeatedly re-discovering or re-committing the same solutions and mistakes, DC enables models to store and reuse accumulated strategies, code snippets, and general problem-solving insights at inference time. This test-time learning enhances performance substantially across a range of tasks without needing explicit ground-truth labels or human feedback. Leveraging DC, Claude 3.5 Sonnet’s accuracy more than doubled on AIME math exams once it began retaining algebraic insights across questions. Similarly, GPT-4o’s success rate on the Game of 24 puzzle increased from about 10% to 99% after the model discovered and reused a Python-based solution. In tasks prone to arithmetic mistakes, such as balancing equations, DC enabled GPT-4o and Claude to reach near-perfect accuracy by recalling previously validated code, whereas their baselines stagnated around 50%. Beyond arithmetic challenges, DC yields notable accuracy gains on knowledge-demanding tasks. Claude achieved a 9% improvement in GPQA-Diamond and an 8% boost on MMLU-Pro Engineering and Physics problems. Crucially, DC’s memory is self-curated, focusing on concise, transferable snippets rather than entire transcripts, thereby facilitating meta-learning and avoiding context ballooning. Unlike fine-tuning or static retrieval methods, DC adapts LMs’ problem-solving skills on the fly, without modifying their underlying parameters, and offers a practical approach for continuously refining responses and cutting routine errors. Overall, our findings present DC as a promising approach for augmenting LMs with persistent memory, bridging the divide between isolated inference events and the cumulative, experience-driven learning characteristic of human cognition.
%U https://aclanthology.org/2026.eacl-long.333/
%P 7080-7106
Markdown (Informal)
[Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory](https://aclanthology.org/2026.eacl-long.333/) (Suzgun et al., EACL 2026)
ACL
- Mirac Suzgun, Mert Yuksekgonul, Federico Bianchi, Dan Jurafsky, and James Zou. 2026. Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7080–7106, Rabat, Morocco. Association for Computational Linguistics.