MetaTool: Facilitating Large Language Models to Master Tools with Meta-task Augmentation

TL;DR

MetaTool introduces a self-supervised, meta-task augmentation approach that enables large language models to better understand and generalize tool usage, significantly improving their performance in tool-based tasks and zero-shot scenarios.

Contribution

The paper proposes MetaTool, a novel self-supervised meta-task augmentation method that enhances LLMs' ability to generalize tool understanding across diverse tasks.

Findings

MetaTool achieves performance comparable to ChatGPT in tool-based planning and chatting.

The approach significantly improves zero-shot generalization to new tasks.

Large-scale instruction tuning with MetaTool enhances tool understanding in open-source LLMs.

Abstract

Utilizing tools with Large Language Models (LLMs) is essential for grounding AI agents in real-world applications. The prevailing approach involves few-shot prompting with demonstrations or fine-tuning with expert annotations. However, mere in-context demonstrations may fail to cover sufficient knowledge for complex tools and tasks. Training on solution paths is also hindered by the high cost of expert annotations and generalizing to new tools. A core challenge of generalizable tool use lies in understanding the "meta", or fundamental natures of tools that are transferable across tasks, such as causality and constraints. In this paper, we present MetaTool, a novel tool learning methodology designed to generalize across any reusable toolset. Our approach incorporates a self-supervised augmentation technique derived from a series of meta-tasks. This involves predicting masked elements in the tool execution process. The self-supervised procedure enables scalable generation of high-quality QA data, which is handy for supervising tool understanding. By incorporating meta-task data into task-oriented training, our method significantly enhances the performance of open-source LLMs, achieving results comparable to ChatGPT in both tool-based planning and chatting scenarios. Through large-scale instruction tuning, the MetaTool model demonstrates impressive zero-shot generalizability on new tasks.