This repository serves as a comprehensive survey and curated collection of research papers on "Tool Learning with Large Language Models (LLMs)". It aims to organize the rapidly growing and fragmented literature in this domain, providing a structured overview for researchers and developers seeking to understand and implement LLM-augmented capabilities. The project offers a systematic review of the benefits and methodologies of tool learning, covering task planning, tool selection, tool calling, and response generation.

How It Works

The project's core contribution is a survey paper that categorizes and analyzes existing research. It explores the "why" behind tool learning by detailing the advantages of tool integration and the paradigm itself, and the "how" by breaking down the tool learning workflow into key stages. The survey also summarizes relevant benchmarks, evaluation methods, challenges, and future directions in the field.

Quick Start & Requirements

This repository is a collection of papers and does not have a direct installation or execution command. The primary resource is the survey paper itself, available via a link in the README.

Highlighted Details

• Comprehensive categorization of tool learning research based on a structured workflow (task planning, tool selection, tool calling, response generation).
• Extensive lists of papers, benchmarks, and other resources related to LLM tool learning.
• Detailed discussion of benefits, challenges, and future research directions in the field.
• Includes a structured table of various benchmarks for evaluating LLM tool usage, covering aspects like API planning, tool selection, and robustness.

Maintenance & Community

The project is maintained by quchangle1 and collaborators. Contributions are welcomed via issues or pull requests. The survey paper has been accepted by Frontiers of Computer Science (FCS).

Licensing & Compatibility

The repository itself does not specify a license. The content is primarily a curated list of research papers, each with its own licensing.

Limitations & Caveats

As a survey and paper collection, the repository does not provide executable code or models. The information is based on published research, and the rapidly evolving nature of LLM tool learning means new developments may not be immediately reflected.