Confucius: Iterative Tool Learning from Introspection Feedback by Easy-to-Difficult Curriculum

TL;DR

This paper introduces Confucius, a novel framework that trains large language models to effectively use complex tools through an easy-to-difficult curriculum and iterative self-instruction, enhancing real-world applicability.

Contribution

The paper presents a multi-stage curriculum learning approach combined with iterative self-instruction from introspective feedback for improved tool learning in LLMs.

Findings

Outperforms tuning-free models like ChatGPT and Claude in real-world scenarios.

Demonstrates effectiveness of curriculum learning in training LLMs for complex tool usage.

Shows superiority over existing tuning-based methods such as GPT4Tools.

Abstract

Augmenting large language models (LLMs) with external tools has emerged as a promising approach to extending the capability of LLMs. Although some works employ open-source LLMs for the tool learning task, most of them are trained in a controlled environment in which LLMs only learn to execute the human-provided tools. However, selecting proper tools from the large toolset is also a crucial ability for the tool learning model to be applied in real-world applications. Existing methods usually directly employ self-instruction methods to train the model, which ignores differences in tool complexity. In this paper, we propose the Confucius, a novel tool learning framework to train LLM to use complicated tools in real-world scenarios, which contains two main phases: (1) We first propose a multi-stage learning method to teach the LLM to use various tools from an easy-to-difficult curriculum; (2) thenceforth, we propose the Iterative Self-instruct from Introspective Feedback (ISIF) to dynamically construct the dataset to improve the ability to use the complicated tool. Extensive experiments conducted on both controlled and real-world settings demonstrate the superiority of our tool learning framework in the real-world application scenarios compared to both tuning-free (e.g. ChatGPT, Claude) and tuning-based baselines (e.g. GPT4Tools).