BriefGPT.xyz
Jun, 2024
工具故障:检测有故障的工具中的静默错误
Tools Fail: Detecting Silent Errors in Faulty Tools
HTML
PDF
Jimin Sun, So Yeon Min, Yingshan Chang, Yonatan Bisk
TL;DR
LLMs使用工具检索知识、在网络上执行任务甚至控制机器人,本文介绍了一个框架来更广泛地研究模型对探测'静默'工具错误的能力,并对如何规划进行反思,以更直接地与模型作为工具的流行用途相一致。我们提供了一个初步的故障恢复方法,在控制计算器设置和实体代理规划方面取得了有希望的结果。
Abstract
tools
have become a mainstay of
llms
, allowing them to retrieve knowledge not in their weights, to perform tasks on the web, and even to control robots. However, most ontologies and surveys of tool-use have assum
→