← 返回卡片

原书出处

Designing Large Language Model Applications

First Edition · 核验版本的摘录快照 · 导入于 2026-11-14T13:54:13.947Z

Guardrails and Verifiers In production environments, mistakes can be catastrophic. Depending on the use case, the agent might need to adhere to strict standards in factuality, safety, accuracy, and many other criteria. Safety is ensured by using guardrails, components that ensure models do not overstep their bounds during the course of their workflows. Some examples of guardrails include toxic language detectors, personally identifiable information (PII)

此预览只包含定位所需的摘录和上下文。

原书定位详情

来源:a19c7933-4eda-4176-aebd-30fe7e1ad292

正文位置:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch10.html

EPUB SHA-256:1b4d2347fec2f949fcddfbaf4fad597a1ae20887a096cfae64c59a619b4781bf