Saved in:
Bibliographic Details
Main Authors: Feng, Zihao, Wang, Xiaoxue, Wu, Bowen, Cao, Hailong, Zhao, Tiejun, Yu, Qun, Wang, Baoxun
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.14718
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • While reinforcement learning (RL) is increasingly used for LLM-based tool learning, its efficiency is often hampered by an overabundance of simple samples that provide diminishing learning value as training progresses. Existing dynamic sampling techniques are ill-suited for the multi-task structure and fine-grained reward mechanisms inherent to tool learning. This paper introduces Dynamic Sampling with Curriculum Learning (DSCL), a framework specifically designed to address this challenge by targeting the unique characteristics of tool learning: its multiple interdependent sub-tasks and multi-valued reward functions. DSCL features two core components: Reward-Based Dynamic Sampling, which uses multi-dimensional reward statistics (mean and variance) to prioritize valuable data, and Task-Based Dynamic Curriculum Learning, which adaptively focuses training on less-mastered sub-tasks. Through extensive experiments, we demonstrate that DSCL significantly improves training efficiency and model performance over strong baselines, achieving a 3.29\% improvement on the BFCLv3 benchmark. Our method provides a tailored solution that effectively leverages the complex reward signals and sub-task dynamics within tool learning to achieve superior results.