知识卡片
AI隔离覆盖增强全过程
训练数据、检索索引、租户适配配置及调用目标都可能携带租户信息,须防止跨租户读取或错误路由。共享模型并不让这些周边数据自动共享;在每次访问点实施策略,才能让租户化响应建立在正确资料上。
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
- GenAI and Multi-Tenancy · 查看原书依据
create vector databases, RAG data, or tenant-focused fine-tuning, I must also introduce the tenant isolation policies and strategies that can ensure that these resources are adequately protected. Figure 16-13 provides a conceptual look at protecting your tenant resources with isolation policies.
参考来源
- 位置:《Building Multi-Tenant SaaS Architectures》第16章《GenAI and Multi-Tenancy》(来源项目 a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Building Multi-Tenant SaaS Architectures/_epub-src/OEBPS/ch16.html)
- 结论依据:原文具体说明训练数据、检索索引、租户适配配置及调用目标都可能携带租户信息,须防止跨租户读取或错误路由;本卡据此归纳机制及适用边界。
- 原始内容:create vector databases, RAG data, or tenant-focused fine-tuning, I must also introduce the tenant isolation policies and strategies that can ensure that these resources are adequately protected. Figure 16-13 provides a conceptual look at protecting your tenant resources with isolation policies.