人工智能
262 篇 · 资讯流仅用3天!千问办公协助国家天文台科研团队打造科研级望远镜仿真系统
日前,千问办公协助国家天文台科研团队打造大口径科研级望远镜数字化仿真系统,耗时仅3天,成本不到千元。依托对真实望远镜组件和观测条件的模拟,Agent可在仿真环境中驱动望远镜进行远程态势感知与观测流程仿真。未来,仿真系统产生的数据,还将用于训
原文 ↗OpenAI:为了不被杀死,Agent 竟学会了靠上下文「转世重生」
从 Compaction 到 Artifactory,再到公网文件服务,6 起异常暴露 Agent 状态跨实例传播的新风险。 作者丨郑佳美 编辑丨岑 峰 过去,大模型任务一旦结束,上下文清空,临时状态随之消失,这一轮 AI 的生命周
原文 ↗开源一站式智能终端 uniTerm v1.9.4 发布:跨工作区广播、SSH Kerberos / Agent 认证等 50 余项更新
uniTerm 是一款轻量级一站式开源终端软件。整合终端、文件传输、远程桌面、数据库客户端、容器等 5 大类协议于一身,覆盖 30+ 协议,安装包仅 14 MB;内置可自主执行的 AI Agent,能规划并执行多轮 Shell 命令。 v1
原文 ↗Large Language Models as Falsifiers for Cyber-Physical Systems
Falsification searches for counterexamples to formal specifications in cyber-physical systems (CPS). With specifications
原文 ↗CoRef-GS: Cooperative Referring Gaussian Splatting for Multi-Agent Scene Understanding
Referring scene understanding for embodied robots requires grounding object- and relation-centric language queries from
原文 ↗Language-model groups overstate consensus when replaying human deliberation on a reasoning task
Full-consensus rates are often treated as indicators of collective cognition, yet depend on how participation and final
原文 ↗Parallelism, critical windows, and separations among diffusion language models
A popular selling point of diffusion large language models (dLLMs) is their capacity for parallelism: the ability to gen
原文 ↗Astronex-World 1.0: Real-Time Interactive World Model Foundation
We present Astronex-World 1.0, an open controllable video world-model foundation. Given a text prompt (text-to-video) or
原文 ↗九识建成首个L4万卡集群,无人驾驶进入多模态大模型新范式
一家做无人车的公司,为什么需要一万多张GPU? 9月10日,九识CEO孔旗在广州宣布战略升级为“城市级物理AI”。一周后的9月17日,CTO庄立首次披露支撑这次升级的底层能力:九识已建成业内首个L4级万卡集群,算力集群总规模近1.5万卡。九
原文 ↗KoNeoBench: A Curated Evaluation Dataset for LLM Understanding of Korean Neologisms
Large language models (LLMs) are typically evaluated on static benchmarks, even though natural language constantly evolv
原文 ↗