知识卡片
会话记忆保存动作与反馈证据
记忆可包括用户请求、工具参数与返回、验证反馈和已完成步骤,为后续恢复提供状态。应区分模型推断、外部观察和用户确认,避免把生成的计划误记成已执行事实。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 外部工具与智能体
原书出处 · 1 处
- · 查看原书依据
Session memory We would like to store logs of the steps undertaken by the agent during current and past sessions. These logs can also be augmented with feedback from verification processes, both automated and human. The agent can then retrieve session logs for better context understanding and as a guide to determine next actions in the current workflow. Let’s explore this with an example: Prompt:
参考来源
- 位置:《Designing Large Language Model Applications》第10章《外部工具与智能体》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch10.html)
- 结论依据:摘录中关于会话记忆保存动作与反馈证据的定义、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:Session memory We would like to store logs of the steps undertaken by the agent during current and past sessions. These logs can also be augmented with feedback from verification processes, both automated and human. The agent can then retrieve session logs for better context understanding and as a guide to determine next actions in the current workflow. Let’s explore this with an example: Prompt: