开启时会模糊预览图，关闭后正常显示
RT @MLB: Game 7 of the 2025 World Series averaged 51 million viewers combined across the United States, Canada, and Japan making the thrill…
📈 Leading Growth @GroqInc 📌 Prev @a16z @HubSpot @TheHustle 💻 Chronically online: https://t.co/AkbwhoTr0K 📘 Wrote https://t.co/w1DBDrOZdI 🎙 Podcast at @sydlis 👇
RT @geoffreylitt: Kinda funny that “neurosymbolic program synthesis” ended up being “AI uses terminal to call typechecker in a loop”
gpu enjoyer at @modal. he/him. ex @full_stack_dl, @weights_biases (acq. @CoreWeave), phd Berkeley @Redwood_Neuro. try https://t.co/SYWVMCazZ3
总结
A coder, road bike rider, server fortune teller, electronic waste collector, co-founder of KCORES, ex-director at IllaSoft, KingsoftOffice, Juejin.
结构化笔记到子Agent
总结
AI聊多了会失去记忆? 给大家整理了个给AI增加记忆的7种工程实现, 包括 1. 上下文压缩 2. RAG + 向量数据库 3. 知识图谱记忆 4. 结构化笔记 5. 分层记忆架构 6. 子代理架构 7. 动态上下文管理 (有些部分其实有交叉, 所以分类并不严格) 建议注意好的实践方式还是先尝试上下文压缩, 毕竟简单见效快, 然后考虑昨天给大家分享的 Anthropic 方法, 使用 MCP 来减少上下文消耗. 这些简单的都尝试完毕了, 再上复杂的框架.
为什么AI需要记忆
RT @garrytan: Wabi is one of the most important new consumer codegen apps — it's a little bit of an Instagram moment for true no-code all o…
AI Apps investing @ A16Z; A1111; Boards of Krea, Deel, Clutch, Titan, Arc Boats, Untitled, Happy Robot + more; If you’re not at the table, you’re on the menu