@inproceedings{kim-etal-2025-behaviorsft,
title = "{B}ehavior{SFT}: Behavioral Token Conditioning for Health Agents Across the Proactivity Spectrum",
author = "Kim, Yubin and
Hu, Zhiyuan and
Jeong, Hyewon and
Park, Eugene W and
Li, Shuyue Stella and
Park, Chanwoo and
Xiong, Shiyun and
Lu, MingYu and
Lee, Hyeonhoon and
Liu, Xin and
McDuff, Daniel and
Breazeal, Cynthia and
Tulebaev, Samir and
Park, Hae Won",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.520/",
pages = "9789--9817",
ISBN = "979-8-89176-335-7",
abstract = "Large Language Models (LLMs) as agents require careful behavioral adaptation. While adept at reactive tasks (e.g., medical reasoning), LLMs often struggle with proactive engagement, like unprompted identification of critical missing information or risks. We introduce **BehaviorBench**, a comprehensive dataset to evaluate agent behaviors across a clinical assistance spectrum. To rigorously test the current models, we also introduce **BehaviorBench-Hard**, a challenging subset where the performance of state-of-the-art models drops significantly, revealing weaknesses. To address these challenges, we propose **BehaviorSFT**, a novel training strategy using behavioral tokens to explicitly condition LLMs for dynamic behavioral selection which boosts performance on both benchmarks. Crucially, a blind clinician evaluation confirmed that our trained agents exhibit more realistic clinical behavior, striking a superior balance between helpful proactivity and necessary restraint versus standard fine-tuning or explicitly instructed agents. Project Page: https://behavior-adaptation.github.io/"
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kim-etal-2025-behaviorsft">
<titleInfo>
<title>BehaviorSFT: Behavioral Token Conditioning for Health Agents Across the Proactivity Spectrum</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yubin</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhiyuan</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyewon</namePart>
<namePart type="family">Jeong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Eugene</namePart>
<namePart type="given">W</namePart>
<namePart type="family">Park</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shuyue</namePart>
<namePart type="given">Stella</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chanwoo</namePart>
<namePart type="family">Park</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shiyun</namePart>
<namePart type="family">Xiong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">MingYu</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyeonhoon</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daniel</namePart>
<namePart type="family">McDuff</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cynthia</namePart>
<namePart type="family">Breazeal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Samir</namePart>
<namePart type="family">Tulebaev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hae</namePart>
<namePart type="given">Won</namePart>
<namePart type="family">Park</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) as agents require careful behavioral adaptation. While adept at reactive tasks (e.g., medical reasoning), LLMs often struggle with proactive engagement, like unprompted identification of critical missing information or risks. We introduce **BehaviorBench**, a comprehensive dataset to evaluate agent behaviors across a clinical assistance spectrum. To rigorously test the current models, we also introduce **BehaviorBench-Hard**, a challenging subset where the performance of state-of-the-art models drops significantly, revealing weaknesses. To address these challenges, we propose **BehaviorSFT**, a novel training strategy using behavioral tokens to explicitly condition LLMs for dynamic behavioral selection which boosts performance on both benchmarks. Crucially, a blind clinician evaluation confirmed that our trained agents exhibit more realistic clinical behavior, striking a superior balance between helpful proactivity and necessary restraint versus standard fine-tuning or explicitly instructed agents. Project Page: https://behavior-adaptation.github.io/</abstract>
<identifier type="citekey">kim-etal-2025-behaviorsft</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.520/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>9789</start>
<end>9817</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T BehaviorSFT: Behavioral Token Conditioning for Health Agents Across the Proactivity Spectrum
%A Kim, Yubin
%A Hu, Zhiyuan
%A Jeong, Hyewon
%A Park, Eugene W.
%A Li, Shuyue Stella
%A Park, Chanwoo
%A Xiong, Shiyun
%A Lu, MingYu
%A Lee, Hyeonhoon
%A Liu, Xin
%A McDuff, Daniel
%A Breazeal, Cynthia
%A Tulebaev, Samir
%A Park, Hae Won
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F kim-etal-2025-behaviorsft
%X Large Language Models (LLMs) as agents require careful behavioral adaptation. While adept at reactive tasks (e.g., medical reasoning), LLMs often struggle with proactive engagement, like unprompted identification of critical missing information or risks. We introduce **BehaviorBench**, a comprehensive dataset to evaluate agent behaviors across a clinical assistance spectrum. To rigorously test the current models, we also introduce **BehaviorBench-Hard**, a challenging subset where the performance of state-of-the-art models drops significantly, revealing weaknesses. To address these challenges, we propose **BehaviorSFT**, a novel training strategy using behavioral tokens to explicitly condition LLMs for dynamic behavioral selection which boosts performance on both benchmarks. Crucially, a blind clinician evaluation confirmed that our trained agents exhibit more realistic clinical behavior, striking a superior balance between helpful proactivity and necessary restraint versus standard fine-tuning or explicitly instructed agents. Project Page: https://behavior-adaptation.github.io/
%U https://aclanthology.org/2025.findings-emnlp.520/
%P 9789-9817
Markdown (Informal)
[BehaviorSFT: Behavioral Token Conditioning for Health Agents Across the Proactivity Spectrum](https://aclanthology.org/2025.findings-emnlp.520/) (Kim et al., Findings 2025)
ACL
- Yubin Kim, Zhiyuan Hu, Hyewon Jeong, Eugene W Park, Shuyue Stella Li, Chanwoo Park, Shiyun Xiong, MingYu Lu, Hyeonhoon Lee, Xin Liu, Daniel McDuff, Cynthia Breazeal, Samir Tulebaev, and Hae Won Park. 2025. BehaviorSFT: Behavioral Token Conditioning for Health Agents Across the Proactivity Spectrum. In Findings of the Association for Computational Linguistics: EMNLP 2025, pages 9789–9817, Suzhou, China. Association for Computational Linguistics.