From Failure to Mastery: Generating Hard Samples for Tool-use Agents
This addresses the need for better training corpora in tool-use agents, offering a novel data generation method that is incremental but impactful for the field.
The paper tackles the problem of generating diverse and complex training data for LLM agents with tool-use capabilities by introducing HardGen, an automatic pipeline that creates hard tool-use samples with verifiable reasoning, resulting in a 4B parameter model trained on their dataset outperforming leading competitors like GPT-5.2 and Claude-Opus-4.5.
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.