@inproceedings{yu-etal-2026-stretch,
title = "{STRETCH} the Boundaries: A Unified Self-Taught Framework for Progressive {LLM} Evolution",
author = "Yu, Yajie and
Lee, Mark and
Feng, Yue",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1168/",
pages = "24732--24747",
ISBN = "979-8-89176-462-0",
abstract = "Large language models (LLMs) often suffer from capability stagnation in self-improvement training due to fixed difficulty levels. To address this issue, we propose STRETCH (Self-Taught Reasoning Evolution via Targeted CHallenge) inspired by cognitive scaffolding theory. STRETCH introduces a dynamic Stretch Zone mechanism that continuously aligns question difficulty with the model{'}s solving capability. Within a single parameter space, the model alternates between a Scaffolder that generates adaptive challenges and a Learner that solves them through reinforcement learning. This dual-loop co-evolution process stabilize training and promote progressive reasoning growth. Experiments on both negotiation and operation research benchmarks demonstrate that STRETCH consistently outperforms strong prompting and domain-specific baselines. Further analysis shows that dynamic difficulty alignment is critical for sustained capability improvement and synchronized reasoning evolution."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="yu-etal-2026-stretch">
<titleInfo>
<title>STRETCH the Boundaries: A Unified Self-Taught Framework for Progressive LLM Evolution</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yajie</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mark</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Feng</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>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Large language models (LLMs) often suffer from capability stagnation in self-improvement training due to fixed difficulty levels. To address this issue, we propose STRETCH (Self-Taught Reasoning Evolution via Targeted CHallenge) inspired by cognitive scaffolding theory. STRETCH introduces a dynamic Stretch Zone mechanism that continuously aligns question difficulty with the model’s solving capability. Within a single parameter space, the model alternates between a Scaffolder that generates adaptive challenges and a Learner that solves them through reinforcement learning. This dual-loop co-evolution process stabilize training and promote progressive reasoning growth. Experiments on both negotiation and operation research benchmarks demonstrate that STRETCH consistently outperforms strong prompting and domain-specific baselines. Further analysis shows that dynamic difficulty alignment is critical for sustained capability improvement and synchronized reasoning evolution.</abstract>
<identifier type="citekey">yu-etal-2026-stretch</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1168/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>24732</start>
<end>24747</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T STRETCH the Boundaries: A Unified Self-Taught Framework for Progressive LLM Evolution
%A Yu, Yajie
%A Lee, Mark
%A Feng, Yue
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F yu-etal-2026-stretch
%X Large language models (LLMs) often suffer from capability stagnation in self-improvement training due to fixed difficulty levels. To address this issue, we propose STRETCH (Self-Taught Reasoning Evolution via Targeted CHallenge) inspired by cognitive scaffolding theory. STRETCH introduces a dynamic Stretch Zone mechanism that continuously aligns question difficulty with the model’s solving capability. Within a single parameter space, the model alternates between a Scaffolder that generates adaptive challenges and a Learner that solves them through reinforcement learning. This dual-loop co-evolution process stabilize training and promote progressive reasoning growth. Experiments on both negotiation and operation research benchmarks demonstrate that STRETCH consistently outperforms strong prompting and domain-specific baselines. Further analysis shows that dynamic difficulty alignment is critical for sustained capability improvement and synchronized reasoning evolution.
%U https://aclanthology.org/2026.findings-emnlp.1168/
%P 24732-24747
Markdown (Informal)
[STRETCH the Boundaries: A Unified Self-Taught Framework for Progressive LLM Evolution](https://aclanthology.org/2026.findings-emnlp.1168/) (Yu et al., Findings 2026)
ACL