library_name: transformers
pipeline_tag: text-generation
base_model: Qwen/Qwen2.5-7B-Instruct
tags:
- tool-use
- agent
- reinforcement-learning
ToolMaster-7B
ToolMaster is a framework that shifts tool learning from static imitation to a trial-and-execution paradigm, enabling Large Language Models (LLMs) to actively master tools. It was introduced in the paper Teaching LLMs to Learn Tool Trialing and Execution through Environment Interaction.
Introduction
Existing tool-use paradigms primarily rely on memorizing static solution paths during training, which limits the ability of LLMs to generalize to new or evolving tools. ToolMaster addresses this by training agents to:
- Trial Phase: Conduct autonomous tool trials to accumulate experiential knowledge.
- Execution Phase: Perform planning and solving while explicitly employing self-correction to rectify errors based on environmental feedback.
By leveraging Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL) using Group Relative Policy Optimization (GRPO), ToolMaster empowers agents to dynamically adapt to unfamiliar tools, significantly enhancing generalization and robustness.
Resources
- Paper: Teaching LLMs to Learn Tool Trialing and Execution through Environment Interaction
- Repository: https://github.com/NEUIR/ToolMaster
Model Details
This checkpoint is a fine-tuned version of Qwen2.5-7B-Instruct. It has been optimized for tool planning and invocation through the trial-and-execution framework.
Usage
For detailed instructions on environment setup, data preparation, and evaluation (on benchmarks like ToolHop, TMDB, and StableToolBench), please refer to the official GitHub repository.
Citation
If you find this work useful, please cite:
@article{gao2025teaching,
title={Teaching LLMs to Learn Tool Trialing and Execution through Environment Interaction},
author={Gao, Xingjie and others},
journal={arXiv preprint arXiv:2601.12762},
year={2025}
}