Self-Challenging Language Model Agents

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhou, Yifei, Levine, Sergey, Weston, Jason, Li, Xian, Sukhbaatar, Sainbayar
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918042958561280
author Zhou, Yifei
Levine, Sergey
Weston, Jason
Li, Xian
Sukhbaatar, Sainbayar
author_facet Zhou, Yifei
Levine, Sergey
Weston, Jason
Li, Xian
Sukhbaatar, Sainbayar
contents Large language models are quickly becoming the foundation for intelligent agents that are capable of using tools. However, training such agents is challenging because it requires human creation and annotation of a diverse set of tasks, tools, and evaluation criteria. In this paper, we propose the Self-Challenging framework for training an agent on high-quality tasks that are generated by itself. The agent first plays the role of challenger and generates a task after interacting with the given tools. The tasks take the form of a novel general class of problems termed Code-as-Task, which are defined by an instruction, a verification function and solution and failure cases which serve as tests, allowing to filter only for high-quality tasks. The agent then takes an executor role and trains on those tasks with reinforcement learning using the evaluation feedback as a reward. Evaluation on two existing multi-turn tool-use agent benchmarks, M3ToolEval and TauBench, shows the Self-Challenging framework achieves over a two-fold improvement in Llama-3.1-8B-Instruct, despite using only self-generated training data.
format Preprint
id arxiv_https___arxiv_org_abs_2506_01716
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Self-Challenging Language Model Agents
Zhou, Yifei
Levine, Sergey
Weston, Jason
Li, Xian
Sukhbaatar, Sainbayar
Artificial Intelligence
Computation and Language
Large language models are quickly becoming the foundation for intelligent agents that are capable of using tools. However, training such agents is challenging because it requires human creation and annotation of a diverse set of tasks, tools, and evaluation criteria. In this paper, we propose the Self-Challenging framework for training an agent on high-quality tasks that are generated by itself. The agent first plays the role of challenger and generates a task after interacting with the given tools. The tasks take the form of a novel general class of problems termed Code-as-Task, which are defined by an instruction, a verification function and solution and failure cases which serve as tests, allowing to filter only for high-quality tasks. The agent then takes an executor role and trains on those tasks with reinforcement learning using the evaluation feedback as a reward. Evaluation on two existing multi-turn tool-use agent benchmarks, M3ToolEval and TauBench, shows the Self-Challenging framework achieves over a two-fold improvement in Llama-3.1-8B-Instruct, despite using only self-generated training data.
title Self-Challenging Language Model Agents
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2506.01716