@inproceedings{feng-etal-2026-adapting,
title = "Adapting Dynamic Sampling to Fine-Grained Rewards for Tool Learning with Curriculum Learning",
author = "Feng, Zihao and
Wang, Xiaoxue and
Wu, Bowen and
Xu, Zhen and
Cao, Hailong and
Zhao, Tiejun and
Wang, Baoxun",
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.986/",
pages = "20834--20853",
ISBN = "979-8-89176-462-0",
abstract = "Dynamic sampling is a technique employed in reinforcement Learning for Large Language Models to mitigate training instability, often caused by the gradient-decreasing problem. However, existing dynamic sampling approaches are predominantly designed for tasks with binary rewards (success/failure) and struggle to adapt to the complex reward structures of tool learning. Such task involves interdependent sub-tasks of varying difficulty and yields fine-grained, multi-faceted rewards. To bridge this gap, we introduce Dynamic Sampling with Curriculum Learning (DSCL), an algorithm tailored for the intricate dynamics of tool learning. DSCL integrates two core components: a Reward-Based Dynamic Sampler, which leverages multi-dimensional reward statistics to prioritize high-value training data; and a Task-Based Dynamic Curriculum Learning method to overcome the credit assignment problem caused by the fine-grained rewards. Extensive experiments on widely used tool learning benchmarks demonstrate the efficacy of our approach. DSCL significantly improves model performance, outperforming strong baselines by 4.75{\%} on the BFCL V3 dataset and 4.02{\%} on the API-Bank dataset."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="feng-etal-2026-adapting">
<titleInfo>
<title>Adapting Dynamic Sampling to Fine-Grained Rewards for Tool Learning with Curriculum Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zihao</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoxue</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bowen</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hailong</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tiejun</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Baoxun</namePart>
<namePart type="family">Wang</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>Dynamic sampling is a technique employed in reinforcement Learning for Large Language Models to mitigate training instability, often caused by the gradient-decreasing problem. However, existing dynamic sampling approaches are predominantly designed for tasks with binary rewards (success/failure) and struggle to adapt to the complex reward structures of tool learning. Such task involves interdependent sub-tasks of varying difficulty and yields fine-grained, multi-faceted rewards. To bridge this gap, we introduce Dynamic Sampling with Curriculum Learning (DSCL), an algorithm tailored for the intricate dynamics of tool learning. DSCL integrates two core components: a Reward-Based Dynamic Sampler, which leverages multi-dimensional reward statistics to prioritize high-value training data; and a Task-Based Dynamic Curriculum Learning method to overcome the credit assignment problem caused by the fine-grained rewards. Extensive experiments on widely used tool learning benchmarks demonstrate the efficacy of our approach. DSCL significantly improves model performance, outperforming strong baselines by 4.75% on the BFCL V3 dataset and 4.02% on the API-Bank dataset.</abstract>
<identifier type="citekey">feng-etal-2026-adapting</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.986/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>20834</start>
<end>20853</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Adapting Dynamic Sampling to Fine-Grained Rewards for Tool Learning with Curriculum Learning
%A Feng, Zihao
%A Wang, Xiaoxue
%A Wu, Bowen
%A Xu, Zhen
%A Cao, Hailong
%A Zhao, Tiejun
%A Wang, Baoxun
%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 feng-etal-2026-adapting
%X Dynamic sampling is a technique employed in reinforcement Learning for Large Language Models to mitigate training instability, often caused by the gradient-decreasing problem. However, existing dynamic sampling approaches are predominantly designed for tasks with binary rewards (success/failure) and struggle to adapt to the complex reward structures of tool learning. Such task involves interdependent sub-tasks of varying difficulty and yields fine-grained, multi-faceted rewards. To bridge this gap, we introduce Dynamic Sampling with Curriculum Learning (DSCL), an algorithm tailored for the intricate dynamics of tool learning. DSCL integrates two core components: a Reward-Based Dynamic Sampler, which leverages multi-dimensional reward statistics to prioritize high-value training data; and a Task-Based Dynamic Curriculum Learning method to overcome the credit assignment problem caused by the fine-grained rewards. Extensive experiments on widely used tool learning benchmarks demonstrate the efficacy of our approach. DSCL significantly improves model performance, outperforming strong baselines by 4.75% on the BFCL V3 dataset and 4.02% on the API-Bank dataset.
%U https://aclanthology.org/2026.findings-emnlp.986/
%P 20834-20853
Markdown (Informal)
[Adapting Dynamic Sampling to Fine-Grained Rewards for Tool Learning with Curriculum Learning](https://aclanthology.org/2026.findings-emnlp.986/) (Feng et al., Findings 2026)
ACL