ToolForge: A Data Synthesis Pipeline for Multi-Hop Search without Real-World APIs

2025-12-17

AI Summary

Existing tool learning approaches rely on real API calls, incurring high computational costs, exhibiting poor generalization, and lacking multi-hop reasoning and self-reflection capabilities. Method: We propose the first real-API-call-free framework for synthesizing multi-hop search tool learning data. Given (question, gold context, answer) triplets, it automatically generates high-quality, diverse training data via lightweight virtual tool modeling. Our method innovatively integrates multi-hop reasoning chain generation with self-reflection enhancement, and establishes a multi-layer verification system—combining rule-based and model-based checks—to ensure data fidelity. Contribution/Results: Experiments demonstrate that an 8B-parameter model trained on our synthetic data surpasses GPT-4o across multiple benchmarks. To foster reproducibility and community advancement, we publicly release both the code and the dataset.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Training LLMs to invoke tools and leverage retrieved information necessitates high-quality, diverse data. However, existing pipelines for synthetic data generation often rely on tens of thousands of real API calls to enhance generalization, incurring prohibitive costs while lacking multi-hop reasoning and self-reflection. To address these limitations, we introduce ToolForge, an automated synthesis framework that achieves strong real-world tool-calling performance by constructing only a small number of virtual tools, eliminating the need for real API calls. ToolForge leverages a (question, golden context, answer) triple to synthesize large-scale tool-learning data specifically designed for multi-hop search scenarios, further enriching the generated data through multi-hop reasoning and self-reflection mechanisms. To ensure data fidelity, we employ a Multi-Layer Validation Framework that integrates both rule-based and model-based assessments. Empirical results show that a model with only 8B parameters, when trained on our synthesized data, outperforms GPT-4o on multiple benchmarks. Our code and dataset are publicly available at https://github.com/Buycar-arb/ToolForge .

Problem

• Synthesizes tool-learning data without real API calls
• Enables multi-hop search with self-reflection mechanisms
• Trains smaller models to outperform larger ones on benchmarks

Innovation

• Automated synthesis framework eliminates real API calls
• Multi-hop reasoning and self-reflection enrich generated data
• Multi-Layer Validation ensures data fidelity with rule and model assessments