AutoDetect: Towards a Unified Framework for Automated Weakness Detection in Large Language Models

TL;DR

AutoDetect introduces a unified, automated framework using LLM-powered agents to systematically identify and address weaknesses in large language models, leading to improved performance and more targeted model enhancements.

Contribution

The paper presents AutoDetect, a novel framework that automates weakness detection in LLMs through collaborative agents, surpassing traditional benchmarking and manual inspection methods.

Findings

Achieves over 30% success rate in identifying flaws in models like ChatGPT and Claude.

Guides targeted improvements, resulting in over 10% performance boost on multiple benchmarks.

Outperforms untargeted data augmentation methods such as Self-Instruct.

Abstract

Although Large Language Models (LLMs) are becoming increasingly powerful, they still exhibit significant but subtle weaknesses, such as mistakes in instruction-following or coding tasks. As these unexpected errors could lead to severe consequences in practical deployments, it is crucial to investigate the limitations within LLMs systematically. Traditional benchmarking approaches cannot thoroughly pinpoint specific model deficiencies, while manual inspections are costly and not scalable. In this paper, we introduce a unified framework, AutoDetect, to automatically expose weaknesses in LLMs across various tasks. Inspired by the educational assessment process that measures students' learning outcomes, AutoDetect consists of three LLM-powered agents: Examiner, Questioner, and Assessor. The collaboration among these three agents is designed to realize comprehensive and in-depth weakness identification. Our framework demonstrates significant success in uncovering flaws, with an identification success rate exceeding 30% in prominent models such as ChatGPT and Claude. More importantly, these identified weaknesses can guide specific model improvements, proving more effective than untargeted data augmentation methods like Self-Instruct. Our approach has led to substantial enhancements in popular LLMs, including the Llama series and Mistral-7b, boosting their performance by over 10% across several benchmarks. Code and data are publicly available at https://github.com/thu-coai/AutoDetect.