@inproceedings{lin-etal-2026-interactive,
title = "Interactive Learning for {LLM} Reasoning",
author = "Lin, Hehai and
Cao, Shilei and
Wang, Sudong and
Wu, Haotian and
Li, Minzhi and
Yang, Linyi and
Zheng, Juepeng and
Qin, Chengwei",
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.303/",
pages = "6085--6108",
ISBN = "979-8-89176-395-1",
abstract = "Existing multi-agent learning approaches explicitly foster collaboration among Large Language Models (LLMs) to build stronger multi-agent systems (MAS), yet they still rely on re-executing the MAS during inference. This contrasts with human cognition, wherein individuals can internalize insights from interactions to improve later independent reasoning. To investigate whether multi-agent interaction can enhance LLMs' independent problem-solving ability, we propose ILR (Interactive Learning for LLM Reasoning), a co-learning framework that integrates Dynamic Interaction and Perception Calibration. Dynamic Interaction adaptively selects cooperative or competitive strategies based on question difficulty and model capability, after which LLMs exchange information via Idea3 framework (Idea Sharing, Idea Analysis, and Idea Fusion), an interaction paradigm simulating human discussion, before producing final answers. Perception Calibration employs Group Relative Policy Optimization (GRPO) while integrating one LLM{'}s reward characteristics into another{'}s to strengthen interaction cohesion. We evaluate the effectiveness of ILR across three LLMs from two model families of varying scales on five mathematical and one coding benchmarks. We further investigate the advantage of Dynamic Interaction (i.e., boosting the robustness of stronger LLMs and surpassing pure strategy), and the scalability of ILR beyond two-model interactions."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="lin-etal-2026-interactive">
<titleInfo>
<title>Interactive Learning for LLM Reasoning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hehai</namePart>
<namePart type="family">Lin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shilei</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sudong</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haotian</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Minzhi</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linyi</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juepeng</namePart>
<namePart type="family">Zheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chengwei</namePart>
<namePart type="family">Qin</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>Existing multi-agent learning approaches explicitly foster collaboration among Large Language Models (LLMs) to build stronger multi-agent systems (MAS), yet they still rely on re-executing the MAS during inference. This contrasts with human cognition, wherein individuals can internalize insights from interactions to improve later independent reasoning. To investigate whether multi-agent interaction can enhance LLMs’ independent problem-solving ability, we propose ILR (Interactive Learning for LLM Reasoning), a co-learning framework that integrates Dynamic Interaction and Perception Calibration. Dynamic Interaction adaptively selects cooperative or competitive strategies based on question difficulty and model capability, after which LLMs exchange information via Idea3 framework (Idea Sharing, Idea Analysis, and Idea Fusion), an interaction paradigm simulating human discussion, before producing final answers. Perception Calibration employs Group Relative Policy Optimization (GRPO) while integrating one LLM’s reward characteristics into another’s to strengthen interaction cohesion. We evaluate the effectiveness of ILR across three LLMs from two model families of varying scales on five mathematical and one coding benchmarks. We further investigate the advantage of Dynamic Interaction (i.e., boosting the robustness of stronger LLMs and surpassing pure strategy), and the scalability of ILR beyond two-model interactions.</abstract>
<identifier type="citekey">lin-etal-2026-interactive</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.303/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>6085</start>
<end>6108</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Interactive Learning for LLM Reasoning
%A Lin, Hehai
%A Cao, Shilei
%A Wang, Sudong
%A Wu, Haotian
%A Li, Minzhi
%A Yang, Linyi
%A Zheng, Juepeng
%A Qin, Chengwei
%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 lin-etal-2026-interactive
%X Existing multi-agent learning approaches explicitly foster collaboration among Large Language Models (LLMs) to build stronger multi-agent systems (MAS), yet they still rely on re-executing the MAS during inference. This contrasts with human cognition, wherein individuals can internalize insights from interactions to improve later independent reasoning. To investigate whether multi-agent interaction can enhance LLMs’ independent problem-solving ability, we propose ILR (Interactive Learning for LLM Reasoning), a co-learning framework that integrates Dynamic Interaction and Perception Calibration. Dynamic Interaction adaptively selects cooperative or competitive strategies based on question difficulty and model capability, after which LLMs exchange information via Idea3 framework (Idea Sharing, Idea Analysis, and Idea Fusion), an interaction paradigm simulating human discussion, before producing final answers. Perception Calibration employs Group Relative Policy Optimization (GRPO) while integrating one LLM’s reward characteristics into another’s to strengthen interaction cohesion. We evaluate the effectiveness of ILR across three LLMs from two model families of varying scales on five mathematical and one coding benchmarks. We further investigate the advantage of Dynamic Interaction (i.e., boosting the robustness of stronger LLMs and surpassing pure strategy), and the scalability of ILR beyond two-model interactions.
%U https://aclanthology.org/2026.findings-acl.303/
%P 6085-6108
Markdown (Informal)
[Interactive Learning for LLM Reasoning](https://aclanthology.org/2026.findings-acl.303/) (Lin et al., Findings 2026)
ACL
- Hehai Lin, Shilei Cao, Sudong Wang, Haotian Wu, Minzhi Li, Linyi Yang, Juepeng Zheng, and Chengwei Qin. 2026. Interactive Learning for LLM Reasoning. In Findings of the Association for Computational Linguistics: ACL 2026, pages 6085–6108, San Diego, California, United States. Association for Computational Linguistics.