BriefGPT - AI 论文速递 ·

Using Large Language Models as Knowledge-Driven Agents for Reliable Retrosynthetic Planning

💡 原文英文，约100词，阅读约需1分钟。

📝

内容提要

本研究提出了一种结合大型语言模型和知识图谱的代理系统，旨在解决聚合物科学中合成路径识别的复杂性。该系统利用多分支反应路径搜索算法，自动化文献检索和数据提取，成功推荐了数百条聚酰亚胺合成路径，展示了其有效性和应用潜力。

🏷️

Presentation: Ontology‐Driven Observability: Building the E2E Knowledge Graph at Netflix Scale
Prasanna Vijayanathan and Renzo Sanchez-Silva share how Netflix tackles obser...
ENPIRE——现实世界中具身智能体机器人策略的自我改进(具身RSI的代表之一)：
ENPIRE提出一种面向编码智能体的自主机器人策略自进化框架，通过两个阶段实现真实世界无干预学习：第一阶段基于人类示范自动构建含安全约束、自动验证与复位机...
YouTube, Meta, and Twitch won’t say if they’ll allow the US government to livestream an execution
YouTube, Twitch, Meta, and X aren't saying whether they'll allow the ...
The AI is in the computer
It's gadget season, and a lot of companies are launching a lot of new pro...
Shopify Upgrades Checkout Blocks to Polaris Web Components, Cutting Bundle Sizes up to 85%
Shopify has updated its Checkout Blocks app, transitioning five high-traffic ...
.NET 真的缺 Raft 组件吗？.NET 生态里长出来的一只「etcd 级」选手 - 张善友
.NET 缺的从来不是库，是一个被顶级基础设施反复用过、踩过坑的共识底座，以及围绕底座长出来的成品服务。而这恰恰是 Kahuna 的价值：它把 .NE...