知识卡片

普通读书笔记卡 · 6765 · 生成模型:GPT-6

对话记忆依赖外部状态组织

理解等级:中级

服务需要把历史或其摘要、检索结果纳入当前请求,才能让回答参考之前内容。模型参数、当前上下文与应用保存的会话是不同状态;保存了历史不等于每次都能正确检索或无限记住,需考虑窗口与选择。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      You really need those semantic capabilities for your LLM to be useful. LLMs in popular applications like ChatGPT, You.com, and Phind use semantic search under the hood. A raw LLM has no memory of anything you’ve said previously. It is completely stateless. You have to give it a run-up to your question every single time you ask it something. For example, when you ask an

    参考来源

    - 位置:《Natural Language Processing in Action, Second Edition》第10章《真实世界中的大语言模型》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-10.html) - 结论依据:摘录中关于对话记忆依赖外部状态组织的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:You really need those semantic capabilities for your LLM to be useful. LLMs in popular applications like ChatGPT, You.com, and Phind use semantic search under the hood. A raw LLM has no memory of anything you’ve said previously. It is completely stateless. You have to give it a run-up to your question every single time you ask it something. For example, when you ask an