@inproceedings{tang-etal-2026-beyond-turn,
title = "Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window",
author = "Tang, Qiaoyu and
Xiang, Hao and
Yu, Le and
Yu, Bowen and
Lu, Yaojie and
Han, Xianpei and
Sun, Le and
Zhang, Wenjuan and
Wang, Pengbo and
Liu, Shixuan and
Zhang, Zhenru and
Tu, Jianhong and
Lin, Hongyu and
Lin, Junyang",
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.2461/",
pages = "52951--52982",
ISBN = "979-8-89176-462-0",
abstract = "While recent advances in reasoning models have demonstrated cognitive behaviors through reinforcement learning, existing approaches struggle to invoke deep reasoning capabilities in multi-turn agents with long-horizon interactions. We propose DeepMiner, a framework that elicits such abilities in deep search agents by introducing a dynamic context window and challenging training tasks. We design a dynamic context management strategy that utilizes a sliding window to selectively drop distant tool responses while preserving all assistant reasoning traces, eliminating the dependency on external summarization models. To provide training tasks that genuinely demand long-horizon reasoning, DeepMiner further employs a reverse construction method to generate complex yet verifiable question-answer pairs from authentic web sources, ensuring both difficulty and reliability. Through reinforcement learning on Qwen3-32B, we develop DeepMiner-32B, which achieves substantial performance improvements across multiple search agent benchmarks. DeepMiner attains 33.5{\%} accuracy on BrowseComp-en, surpassing the much larger DeepSeek-V3.1-671B, and demonstrates consistent improvements on BrowseComp-zh, XBench-DS, and GAIA. Notably, our dynamic context management enables sustained interactions of nearly 100 turns within commonly used 32k context length, effectively addressing the context limitations that constrain existing multi-turn agents."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tang-etal-2026-beyond-turn">
<titleInfo>
<title>Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window</title>
</titleInfo>
<name type="personal">
<namePart type="given">Qiaoyu</namePart>
<namePart type="family">Tang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hao</namePart>
<namePart type="family">Xiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Le</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bowen</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yaojie</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xianpei</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Le</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wenjuan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Pengbo</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shixuan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhenru</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jianhong</namePart>
<namePart type="family">Tu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hongyu</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junyang</namePart>
<namePart type="family">Lin</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>While recent advances in reasoning models have demonstrated cognitive behaviors through reinforcement learning, existing approaches struggle to invoke deep reasoning capabilities in multi-turn agents with long-horizon interactions. We propose DeepMiner, a framework that elicits such abilities in deep search agents by introducing a dynamic context window and challenging training tasks. We design a dynamic context management strategy that utilizes a sliding window to selectively drop distant tool responses while preserving all assistant reasoning traces, eliminating the dependency on external summarization models. To provide training tasks that genuinely demand long-horizon reasoning, DeepMiner further employs a reverse construction method to generate complex yet verifiable question-answer pairs from authentic web sources, ensuring both difficulty and reliability. Through reinforcement learning on Qwen3-32B, we develop DeepMiner-32B, which achieves substantial performance improvements across multiple search agent benchmarks. DeepMiner attains 33.5% accuracy on BrowseComp-en, surpassing the much larger DeepSeek-V3.1-671B, and demonstrates consistent improvements on BrowseComp-zh, XBench-DS, and GAIA. Notably, our dynamic context management enables sustained interactions of nearly 100 turns within commonly used 32k context length, effectively addressing the context limitations that constrain existing multi-turn agents.</abstract>
<identifier type="citekey">tang-etal-2026-beyond-turn</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2461/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>52951</start>
<end>52982</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window
%A Tang, Qiaoyu
%A Xiang, Hao
%A Yu, Le
%A Yu, Bowen
%A Lu, Yaojie
%A Han, Xianpei
%A Sun, Le
%A Zhang, Wenjuan
%A Wang, Pengbo
%A Liu, Shixuan
%A Zhang, Zhenru
%A Tu, Jianhong
%A Lin, Hongyu
%A Lin, Junyang
%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 tang-etal-2026-beyond-turn
%X While recent advances in reasoning models have demonstrated cognitive behaviors through reinforcement learning, existing approaches struggle to invoke deep reasoning capabilities in multi-turn agents with long-horizon interactions. We propose DeepMiner, a framework that elicits such abilities in deep search agents by introducing a dynamic context window and challenging training tasks. We design a dynamic context management strategy that utilizes a sliding window to selectively drop distant tool responses while preserving all assistant reasoning traces, eliminating the dependency on external summarization models. To provide training tasks that genuinely demand long-horizon reasoning, DeepMiner further employs a reverse construction method to generate complex yet verifiable question-answer pairs from authentic web sources, ensuring both difficulty and reliability. Through reinforcement learning on Qwen3-32B, we develop DeepMiner-32B, which achieves substantial performance improvements across multiple search agent benchmarks. DeepMiner attains 33.5% accuracy on BrowseComp-en, surpassing the much larger DeepSeek-V3.1-671B, and demonstrates consistent improvements on BrowseComp-zh, XBench-DS, and GAIA. Notably, our dynamic context management enables sustained interactions of nearly 100 turns within commonly used 32k context length, effectively addressing the context limitations that constrain existing multi-turn agents.
%U https://aclanthology.org/2026.findings-emnlp.2461/
%P 52951-52982
Markdown (Informal)
[Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window](https://aclanthology.org/2026.findings-emnlp.2461/) (Tang et al., Findings 2026)
ACL
- Qiaoyu Tang, Hao Xiang, Le Yu, Bowen Yu, Yaojie Lu, Xianpei Han, Le Sun, Wenjuan Zhang, Pengbo Wang, Shixuan Liu, Zhenru Zhang, Jianhong Tu, Hongyu Lin, and Junyang Lin. 2026. Beyond Turn Limits: Training Deep Search Agents with Dynamic Context Window. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 52951–52982, Budapest, Hungary. Association for Computational Linguistics.