From Failure to Mastery: Generating Hard Samples for Tool-use Agents

TL;DR

This paper introduces HardGen, a novel pipeline for generating challenging, verifiable training samples for tool-use agents, significantly improving their ability to handle complex reasoning tasks.

Contribution

HardGen creates a dynamic, failure-based API graph to synthesize hard training traces, guiding the instantiation of advanced tools for complex reasoning.

Findings

A 4B parameter model trained with HardGen data outperforms leading competitors.

HardGen generates verifiable, complex Chain-of-Thought reasoning samples.

The approach enhances the diversity and difficulty of training data for tool-use agents.

Abstract

The advancement of LLM agents with tool-use capabilities requires diverse and complex training corpora. Existing data generation methods, which predominantly follow a paradigm of random sampling and shallow generation, often yield simple and homogeneous trajectories that fail to capture complex, implicit logical dependencies. To bridge this gap, we introduce HardGen, an automatic agentic pipeline designed to generate hard tool-use training samples with verifiable reasoning. Firstly, HardGen establishes a dynamic API Graph built upon agent failure cases, from which it samples to synthesize hard traces. Secondly, these traces serve as conditional priors to guide the instantiation of modular, abstract advanced tools, which are subsequently leveraged to formulate hard queries. Finally, the advanced tools and hard queries enable the generation of verifiable complex Chain-of-Thought (CoT), with a closed-loop evaluation feedback steering the continuous refinement of the process. Extensive evaluations demonstrate that a 4B parameter model trained with our curated dataset achieves superior performance compared to several leading open-source and closed-source competitors (e.g., GPT-5.2, Gemini-3-Pro and Claude-Opus-4.5). Our code, models, and dataset will be open-sourced to facilitate future research.