知识卡片
检索增强补充任务上下文
RAG在生成前检索数据库或文档,把有权限获得的最新相关材料随问题提供给模型,弥补参数中没有的个人或时效信息。它不必把这些数据写入模型权重,却仍会把内容送进推理链路;检索权限、日志、提供方留存与提示注入都需单独治理,不能从未训练进权重推出隐私自动得到保护。
这张卡片看不懂?
这张卡片的反馈已满 10 条,暂时无法继续提交。
已有反馈 · 0 / 10 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- Chapter 15. AI Inference and Serving · 查看原书依据
the user’s prompt is augmented with traditional queries to provide the necessary context to correctly deliver an answer to the user’s input. Consider, for example, the query: Prompt: Who is the closest friend to my current location? It is clear that a generic LLM has no knowledge
参考来源
- 位置:《Designing Distributed Systems, 2nd Edition》第15章《Chapter 15. AI Inference and Serving》(来源项目 book-sources;源文件:extracted/Designing Distributed Systems, 2nd Edition/_epub-src/OEBPS/ch15.html)
- 结论依据:原文朋友位置例子说明检索与提示组合;隐私结论限定到数据流真实边界。
- 原始内容:the user’s prompt is augmented with traditional queries to provide the necessary context to correctly deliver an answer to the user’s input. Consider, for example, the query: Prompt: Who is the closest friend to my current location? It is clear that a generic LLM has no knowledge