Jev-Mem 论文提出新智能体记忆架构，构建速度提升 6.6 倍
Banger report on building faster memory for AI Agents. If your agent's memory layer is slow, this d...
搞 Agent 记忆层的可以看看这篇：Jev-Mem 用轻量控制器替代 LLM 做决策，构建快 6.6 倍，LoCoMo 上还涨了 11%。
一篇新报告提出了 Jev-Mem，一种受 System-One/System-Two 认知启发的智能体记忆架构。轻量级控制器负责存储时的记忆类型与关系分配，以及读取时的路由、检索预算、图遍历、评分和停止判断，LLM 只用于最终推理和答案生成。在 LoCoMo 基准上，Jev-Mem 以 LLM judge 评分达到 0.777，比最强基线相对提升 11.0%。记忆构建耗时 158 秒，速度提升 6.6 倍，平均查询延迟降至 0.93 秒，降低 36.7%。
Banger report on building faster memory for AI Agents. If your agent's memory layer is slow, this d...
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… 💬 15 🔄 7 ❤️ 76 👀 6645 📊 37 ⚡
