OpenVLHarness: Real World Visual Agentic Tool-Use with Multimodal Memory· OpenVLHarness: 多模态记忆的视觉代理工具使用
Multimodal large language models (MLLMs) have made rapid progress on visual reasoning, yet…
Multimodal large language models (MLLMs) have made rapid progress on visual reasoning, yet…
A seemingly reasonable push in Sokoban can make the puzzle unsolvable. To avoid such mista…
The goal of a world model is to capture the dynamics of an environment. This has the great…
Systems that combine several large language models (LLMs) often use one context for many q…
LLM agents that compute the right answer can still answer wrongly. Agent harnesses increas…
Many modern language models have adopted hybrid architectures that incorporate both linear…
Frontier agents retrieve structural metadata well and compute geometry from coordinates le…
Agent learning is shifting from learning from human data toward learning through interacti…
Context compression improves long-context efficiency but may introduce faithfulness issues…
Beyond basic tool use and coding skills, effectively incorporating visual references into …
Reasoning models often run into problems without noticing them, even when symptoms are vis…
Traditional evaluation benchmarks reduce inherently interconnected scientific knowledge in…
A key problem in AI safety is eliciting latent knowledge (ELK), extracting information fro…
Evaluating an agent's planning ability remains challenging for open-domain tasks, where ve…
AI agents deployed inside organizations can pose an insider threat if they make mistakes, …
LLM trading agents are evaluated almost exclusively in backtests and autonomous benchmark …
Smartphone GUI agents complete tasks in mobile apps by repeatedly reading the screen and a…
An LLM-generated repair action specifies which records to change and by how much. A failed…
Code generation models are increasingly used to produce interactive artifacts, such as web…
Agent harnesses decide what a language model sees across turns: they drop old turns, summa…
Recent time series foundation models (TSFMs) can consume covariates in-context, extending …
Robust partially observable Markov decision processes model agents that act under partial …
Long-horizon embodied agents face temporal confusion and incomplete or distorted records, …
One of the central concerns for the safety of advanced AI is scheming: when a model strate…
Participating in a data space still requires the Dataspace Protocol (DSP) negotiation flow…
I gave Claude Code 120 pages of my handwritten college notes. It gave me back my handwriti…
Local-first AI trend radar and live-session distillation
More than 50% of managers say they're finding it harder to evaluate employees' work after …
From local businesses to global brands, more advertisers can now reach people as they expl…
OpenAI and Google Cloud are bringing frontier models and Codex workflows to Gemini’s Model…
医疗AI质变时刻
At GTC Berlin, NVIDIA is showcasing its full lineup of local AI hardware. RTX Spark, which…
General availability of the instances is coming next month, with planned expansion of avai…
Demand from AI-natives is pulling the European AI cloud ecosystem forward faster than ever…
To increase productivity, accelerate innovation and create new opportunities for economic …
NVIDIA cuDSS, launched today in general availability, is an NVIDIA CUDA-X library that shi…
NVIDIA builds tools to enable safe agents across the stack. The NVIDIA OpenShell Reference…
10月20日,“2026福布斯中国最佳雇主”榜单公布,海信集团荣获2026中国年度最佳雇主!这是海信连续第三年获得福布斯中国雇主系列奖项。从ESG实践到数智化转型,再到综合雇主价值…
Agents are only as useful as what they know, and only as trustworthy as the boundaries the…
NVIDIA is expanding its family of open models across language, robotics, biology, and quan…
10月21日,生数科技世界模型团队正式开启Motubrain 2.0训练全流程公开直播
Founded in a tent near the Berlin Wall in 1980, the Tempodrom moved several times before s…
根据国际数据公司(IDC)最新发布的《中国汽车云市场(2026年上半年)跟踪》报告,2026年上半年中国汽车公共云基础设施市场规模约83.4亿元人民币。其中,阿里云市场份额达45.…
世界模型赛道新看点:千诀隐变量可辨识性方向的研究,比LeCun相近工作早一年,不堆画面预测,主打物理规律与跨本体复用。
Learn how OpenAI built ChatGPT’s LLM-powered ad stack to understand conversational intent,…
机器人在哪里练,练成后用在哪里?越来越多的探索指向同一个答案:把训练放进真实的工位,这就是“实景实训平台”。据中国信息通信研究院《具身智能训练场研究报告(2026年)》及多家机构测…
百万台搭载打底,智驾普及与安全并重
质检线上,一台人形机器人从料框里取出平板,送进检测柜;完成检测后,再将成品分流。抓取入柜、多项目质检、成品分拣,机器人与工人同线作 业,一气呵 成</spa
智能体也开始领“工号”了。8月31日,张江科学之门模力社区, 智能体身份码节点管理平台 <span style="font-family: 微软雅黑, "Microsoft YaH…
See how OpenAI, PerturbAI, and Scripps Research used distributed AI agents to uncover prom…
电池开发,分工远比想象复杂
<img data-src="https://static.leiphone.com/uploads/new/images/20261021/6ad839630cd72.jpg" …
Widespread AI adoption is propelling Southeast Asia's digital economy to a projected $368 …
<section data-mpa-powered-by="yiban.io" data-pm-slice="0 0 []" style="-webkit-tap-highligh…
先看真实问题解决力
The proposed policy seems to offer protections against AI run amok but actually forbids fe…
Before you share the Netflix documentary "AI Gone Rogue" take note. Clem Delangue, Hugging…
Anthropic is reportedly struggling to secure enough compute from cloud providers that incr…
SpaceX hit profitably in the third-quarter, with net income of $4.4 billion. AI was the en…
OpenAI says it's adding Orchestra's team and technology to work on Rosalind, its AI model …
Google has joined Amazon in releasing a developer tool that makes it easy for PyTorch user…
The company says the feature is designed for situations where users might be discussing so…
The former Google CEO owns about 1% stake in the leading AI lab, Bloomberg reported.
Meta's AI companion Muse has surpassed 10 million U.S. downloads in its first 40 days, joi…
It’s the moment everyone is waiting for — the ChatGPT moment for robotics, Amulya Vishwana…
Electric Plant has built an AI-powered translator to help humans listen to what plants are…
来源:CVer 文章链接:https://mp.weixin.qq.com/s/Xax4gmcHdwmTP
把记忆原生焊进Transformer
Nebius and Nscale are among global technology partners looking to NVIDIA Vera CPUs to buil…
An independent Schlotzsky’s and Cinnabon franchisee cut menu repricing time by 95% and bui…
近日,国际权威咨询公司Gartner正式发布2026年度《容器管理魔力象限》报告,阿里云连续4年入选该报告,稳居“领导者”象限,是亚太地区唯一获此殊荣的科技公司。 Gartner在…
最新发布TokenFleet调度框架
Terminal X’s Eval Agent grades outputs against a 54-criterion rubric, catching 25% more de…
今年4月,在北京的一场影视行业论坛上,爱奇艺基础架构与智能分发事业群总裁刘文峰正通过大屏幕演示手机端的一个新功能“跳看”:前一秒还是画师在寺庙中留下一幅未完成点睛的画作,手指一滑,…
“我73岁转型,我要去拥抱新的技术、去驾驭它,做到真正想做的东西,这就是我的决心。”——鲍德熹 十一月的北京已略带寒意,我们在爱奇艺青春中心见到了一代传奇影人鲍德熹。这里是他和爱奇…
<img data-src="https://static.leiphone.com/uploads/new/images/20261020/6ad6e86dbb744.jpg" …
OpenAI is acquiring Orchestra Bio to bring AI-powered operations to life sciences R&D, hel…

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · OpenAI

YouTube · NVIDIA

YouTube · NVIDIA

YouTube · NVIDIA