智能体
x.com ·
@omarsar0
Jev-Mem：面向AI Agent的更快记忆架构
一篇报告提出新的Agent记忆架构Jev-Mem，灵感来自System-One/System-Two认知模式。系统由轻量控制器做记忆决策：存储时分配记忆类型与关系，读取时负责路由、检索预算、图遍历、打分与停止，LLM只用于最终推理和答案撰写。报告称其记忆构建速度提升6.6倍、查询延迟降低36.7%，在LoCoMo上LLM judge总分0.777，相对最强基线提升11.0%，记忆构建耗时158秒，平均查询延迟0.93秒。
AI推荐理由 该帖推荐一篇关于Agent记忆架构的报告，提出受System-One/System-Two认知启发的Jev-Mem架构：由轻量控制器负责记忆类型与关系分配、路由、检索预算、图遍历、打分与停止，LLM仅用于最终推理和答案生成。给出具体指标：记忆构建6.6倍加速、查询延迟降低36.7%、LoCoMo上LLM judge总分0.777、相对最强基线提升11.0%、记忆构建158秒、平均查询延迟0.93秒，并附论文链接。对Agent记忆层性能优化有参考价值。局限：数据来自论文自述，尚未见独立复现；帖子本身为推荐性转述，未提供方法细节。
74
浏览 6,645
点赞 76
评论 15
转发 7
正文
Banger report on building faster memory for AI Agents.
If your agent's memory layer is slow, this design is worth a look.
Great weekend read.
It proposes a new agentic memory architecture, Jev-Mem, inspired by System-One/System-Two cognition.
This builds agent memory 6.6x faster and cuts query latency by 36.7% while scoring higher on LoCoMo.
In this system, a lightweight controller makes the memory decisions.
It assigns memory types and relations when storing, and handles routing, retrieval budget, graph traversal, scoring and stopping when reading. The LLM is called only for final reasoning and answer writing.
On LoCoMo, it scores 0.777 overall with an LLM judge, an 11.0% relative improvement over the strongest baseline. Memory construction takes 158 seconds, and average query latency drops to 0.93 seconds.
Paper: [academy.dair.ai/papers/jev-mem…](https://academy.dair.ai/papers/jev-mem-system-one-controlled-agentic-memory-for-efficient-ai-agents-2609.23986)
