AI 情报文章归档
浏览 AI圈报 已保存的 38270 条 AI 情报文章,覆盖模型、产品、行业、论文、教程与观点方法六大分类。
已收录文章38270
正式分类6
精选内容4775
全部文章
第 352 / 957 页 · 每页 40 条
论文研究普通
Enhancing Diffusion Language Models with Autoregressive Post-Training Weights
Diffusion language models (dLLMs) have emerged as a promising alternative to autoregressive (AR) language models, offering flexible token-update orders and parallel decoding. Recent dLLMs are often initialized from pretrained AR models bef…
信息来源:arXiv
论文研究普通
Exploiting Acoustic and Content-Oriented Speaker Verification Attacks Against Multilingual Voice Anonymization
Attacker ASV systems for voice anonymization have been studied primarily in English, leaving their behavior in multilingual settings largely unexplored. Conventional ASV has shown that both acoustic and contextual information are important…
信息来源:arXiv
论文研究普通
ChartBmkAgent: Harness-Governed Multi-Agent Construction of Chart QA Benchmarks from Sparse Error-Taxonomy Specifications
Multimodal large language models (MLLMs) advance rapidly, while conventional benchmark development lags behind, delaying investigation of newly observed capability gaps. Such investigation requires an expressive task format and an on-deman…
信息来源:arXiv
论文研究普通
DSV-Mem: Evaluating Multimodal Memory in Professional Workflows for MLLM Agents
Conversational MLLM agents are increasingly expected to assist in professional workflows, from AI research and engineering design to product management and business operations. Yet this capability remains underexplored: existing benchmarks…
信息来源:arXiv
论文研究普通
Beyond Corrected Memory: Execution Consistency in Multi-Agent Systems
Shared memory coordinates agents' actions, but correct records do not establish that those actions satisfy task requirements. Memory governance and failure diagnosis regulate or inspect recorded information; they do not by themselves estab…
信息来源:arXiv
论文研究普通
Surviving the Router: Optimizing Skill Injections for Retrieval and Execution
AI agents increasingly rely on modular third-party "skills" that are dynamically selected by skill routers to execute complex tasks. While recent studies highlight the threat of prompt injections embedded in these skills, existing evaluati…
信息来源:arXiv
论文研究普通
When Tools Lie: Reliability of Mathematical Agents Under Corrupted Tool Feedback
Mathematical problem solving often requires deterministic computational steps that agents delegate to tools and implicitly trust. Yet tools can fail silently, returning plausible but incorrect results. How well can agents detect and correc…
信息来源:arXiv
观点 / 方法普通
前 Anthropic 研究员 Jacob Coxon 在纽约市议会听证会上谈自动化 AI 研究时间表
Jacob Coxon(数日前从 Anthropic 离职的研究员)在纽约市议会听证会上表示,自动化 AI 研究已是各大公司的首要目标;他在 OpenAI 任职时曾为自动化 AI 研究员设定约 2027 或 2028 年的里程碑,目前进度达到甚至超出预期,距离实现是以月计而非数年。
信息来源:X:Rohan Paul (@rohanpaul_ai) · OpenAI / Claude
论文研究普通
Natural Language Questions as an Interface for Knowledge Graphs: QRAKEN Graph Distillation and Semantic Self-Healing
Natural-language access to RDF knowledge graphs is a core Semantic Web ambition. Large language models (LLMs) have advanced Text-to-SPARQL, yet on unfamiliar graphs they often generate valid queries that misrepresent the populated data mod…
信息来源:arXiv · GPT
论文研究普通
SAGE: Semantic Anchor-Guided Evolution for Grounded Medical QA Data Synthesis
Developing reliable models for clinical tasks, such as Medical Question Answering (QA), is severely constrained by the limited availability of high-quality, expert-annotated training data. This challenge is exacerbated by stringent privacy…
信息来源:arXiv
论文研究普通
Explainable Rule Mining of IPv6 Extension-Header Presence Patterns from Paired-Vantage Captures
IPv6 extension headers (EHs), such as fragmentation, segment routing, and in-situ telemetry, are operationally important yetwidely dropped in transit, and characterising their behaviour from packet captures is a recurring measurement probl…
信息来源:arXiv
教程 / 实战普通
Green Tests, Broken Architecture: Four Models Review TypeScript Diffs
This is a submission for the Kaggle Benchmarking Challenge. In ProgressCut, it is possible to write...
信息来源:DEV Community
教程 / 实战普通
Can AI Replace Cloud Engineers? I Put It to the Test
The result first I tested how much of my cloud engineering work I could delegate to AI in...
信息来源:DEV Community
教程 / 实战普通
可灵 Kling 4.0 如何制作全网曝光 500 万+的 AI 短片《The Beat》
可灵 AI 拆解 AI 短片《The Beat》的制作过程,该片全网曝光超 500 万。Kling 4.0 支持单次原生生成最长 30 秒视频、单次任务最多 15 项参考素材、最多 10 张关键帧输入,并提升了大动态运镜稳定性与口型匹配。Kling 4.0 Flash 已向黑金年卡会员开放抢先体验,Kling 4.0 将于 10 月正式上线。
信息来源:公众号:可灵AI(快手·视频)
教程 / 实战普通
Dharohar: an open-source AI audio guide that makes you put your phone in your pocket 🌿
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass ...
信息来源:DEV Community
行业动态普通
意大利反垄断机构对 AI 音乐平台 Suno 展开调查:服务条款涉单方面变更订阅费与放弃著作人身权
意大利反垄断机构对 AI 音乐平台 Suno 展开调查,指其服务条款可能让用户权利义务严重失衡。监管机构称 Suno 可单方面变更合同条款、服务功能及订阅资费,且用户被要求让渡的授权范围过宽,甚至包含放弃著作人身权,涉嫌抵触意大利版权法。条款还强制用户通过美国仲裁解决争议、放弃集体诉讼,并指定马萨诸塞州为唯一管辖地。
信息来源:IT之家·人工智能
教程 / 实战普通
Running Hermes Agent on Kubernetes: What Breaks, What Doesn't, and a Production-Safe Setup
Hermes Agent is Nous Research's self-improving AI agent. It builds skills from experience, keeps...
信息来源:DEV Community
教程 / 实战普通
The live realtor model passed. The goodbye failed. published: true
This is a submission for the Kaggle Benchmarking Challenge What I Benchmarked I run a...
信息来源:DEV Community
教程 / 实战普通
WILDCASE: The World Is the Case File — An Outdoor Mystery Game That Gets You Off Your Screen
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass ...
信息来源:DEV Community
论文研究普通
OpenAI 与 Ironclad 合作训练 GPT-6 Astra,在合同工作流任务上平均得分比 GPT-5.6 Sol 高 32%
OpenAI 与合同软件公司 Ironclad 合作,将其专业合同工作流转化为 11 个训练与评估任务,GPT-6 Astra 成为首个用这些任务训练的前沿模型。在研究评估中,Astra 平均得分为 55.0%,GPT-5.6 Sol 为 41.6%,估算每次尝试平均用时从 37.0 分钟降至 19.2 分钟;训练使用基于 SEC EDGAR 公开合同的合成任务和强化学习,未使用客户数据。
信息来源:OpenAI:官网动态(RSS · 排除企业/客户案例) · OpenAI / GPT
行业动态普通
Valve 确认:Linux 版 Steam 客户端目前不支持 Arm 芯片
Valve 明确表示 Steam for Linux 目前不支持基于 Arm 的硬件,此前一名 AYN Odin 3 掌机用户在高通骁龙设备上运行原生 ARM64 版 Steam 客户端并提交错误报告,官方关闭报告后作出上述回应。Valve 现行 Linux 版 Steam 运行环境要求必须搭配 x86_64 架构内核及处理器,ARM64 版 Steam 虽已存在但仅面向 Steam Frame。
信息来源:IT之家·人工智能
观点 / 方法普通
《逃离塔科夫》1.2.0.0 版本更新上线,PvE 模式新增威望系统
Battlestate Games 为《逃离塔科夫》推送 1.2.0.0 版本更新,PvE 模式终于加入威望(Prestige)系统,玩家满足前置条件后可重置角色并获得丰厚奖励。本次更新还调整了战利品池、AI 控制的 PMC 行为及升级机制,部分地点的机器人等级会与所在区域匹配,并改进了个人资料自定义功能,玩家可选择在资料中显示任一结局。
信息来源:IT之家·人工智能
教程 / 实战普通
Forget Claude Code. All you need is Qwen3.8-Flash-Next running locally for agentic coding
Agentic coding on a Strix Halo laptop with Qwen3.8-Flash-Next. Going head to head with Opus 5.5
信息来源:DEV Community · Qwen / Claude
行业动态普通
华为证实与高通专利许可涵盖LogicFolding芯片技术,高通否认
华为与高通10月5日宣布达成多年期广泛专利交叉许可协议,覆盖5G、计算、人工智能、网络等领域,高通还将收购华为部分美国专利。彭博社报道称协议涉及华为逻辑折叠芯片技术,华为向深视新闻证实授权涵盖近封装光学(NPO)互联及LogicFolding相关专利,但高通回应称该说法不属实,且称其为净支付方的报道不准确;华为和高通新闻稿均未提及相关说法。
信息来源:IT之家·人工智能
教程 / 实战普通
Python Workers GA + 1M-token context, routing fixes
This week brought a mix of genuine production upgrades and a security disclosure you need to action...
信息来源:DEV Community
教程 / 实战普通
What AI agents actually do with your MCP server
An MCP server can look healthy in every log and still be hard for agents to use. Requests come in,...
信息来源:DEV Community
教程 / 实战普通
I stopped picking between Claude Code, Codex and Gemini, so I built a shell that picks for me
I pay for more than one AI coding tool. Most of you probably do too. And every day I hit the same...
信息来源:DEV Community · Claude / Gemini
产品发布 / 更新普通
opencode 免费期延长数日
Space Bunny 的免费期即将结束,不过我们会在免费层级中再赞助几天 模型正在调整中,之后才会揭晓
信息来源:X:opencode (@opencode)
教程 / 实战普通
Making a language look like TypeScript made AI hallucination harder to catch
The original developer-first playbook was world-class docs, beginner-friendly SDKs, and advocates at...
信息来源:DEV Community
教程 / 实战普通
How I Tune RAG Pipelines with RAG-LCC: A Hands-On Local Guide
Most RAG frameworks tell you whether an answer was generated. RAG-LCC tries to show you why that...
信息来源:DEV Community
教程 / 实战普通
She used Claude as a diary. The terms of service are now part of the charge.
Florida law requires a threat 'viewable by another person'. A chatbot's review policy supplied that person. What the record shows — and what builders should change.
信息来源:DEV Community · Claude
产品发布 / 更新普通
Google Docs 和 Google Drive 引入 Markdown 原生支持,可保留 .md 格式协作编辑
据 9To5Google 报道,谷歌为 Google Docs 和 Google Drive 加入 Markdown 原生支持,可直接打开、编辑和使用 .md 文件,渲染后的内容含可点击链接和结构化表格,并支持实时编辑和评论。谷歌表示该功能可让 AI 智能体协作编辑文档,并与 Gemini Notebook 同步。功能已于 10 月 5 日开始推送,部分用户可能需等待最多 15 天。
信息来源:IT之家·人工智能 · Gemini
行业动态普通
LG 电子为雷诺首款量产商用 SDV New Trafic E-Tech Electric 提供集成座舱解决方案
LG 电子宣布为雷诺集团首款量产商用 SDV——New Trafic E-Tech Electric 货车提供集成座舱解决方案,该方案结合数字仪表板与信息娱乐系统,构建针对 SDV 优化的控制框架,实现更直观统一的 UX/UI,并支持以软件为中心的车辆开发方式,帮助车企通过零件简化和设计优化降低生产成本。LG 电子计划进一步扩大与全球主要汽车制造商的合作。
信息来源:IT之家·人工智能
教程 / 实战普通
Fan-Out: Concurrent Tool Calls Under Shared-State Hazards
DEV × Kaggle Benchmarking Challenge · #kagglechallenge Author: Hotragn Pettugani · Kaggle ...
信息来源:DEV Community
产品发布 / 更新普通
AIHOT 上线模型性价比帕累托前沿功能
AIHOT 上线帕累托前沿功能,可查看模型性价比,切换到右上角"性价比"即可访问。不同场景采用不同 Token 权重算法:综合与专业办公按缓存命中:输入:输出=7:2:1,编程按 97:2:1,科研按输入:输出=1:3,知识问答按 3:1。
信息来源:X:卡兹克 (@Khazix0918)
教程 / 实战普通
Don't put an LLM where a bash script will do
Don't put an LLM where a bash script will do Or: how I almost gave a language model a...
信息来源:DEV Community
教程 / 实战普通
The scarcest skill on my team has the lowest status: the 'no'
There's a person on my team everyone quietly files as "the negative one." He's the one who, when a...
信息来源:DEV Community
行业动态普通
波兰反垄断机构 UOKiK 指控谷歌在与出版商报酬谈判中滥用市场支配地位,最高可罚营业额 10%
波兰竞争与消费者保护办公室 UOKiK 指控 Alphabet 及谷歌多家实体在与波兰媒体出版商协商内容报酬时滥用市场支配地位,涉及谷歌搜索、谷歌新闻和 Discover 中展示新闻与摘要的报酬拟定方式。
信息来源:IT之家·人工智能
行业动态普通
月之暗面拟 2027 年一季度赴港 IPO,可灵 AI 也在筹备香港上市
据 Bloomberg 援引知情人士报道,Kimi 开发商月之暗面(Moonshot AI)计划 2027 年一季度在香港 IPO,此前一轮私募融资估值约 $50 billion。快手的 AI 视频平台可灵(Kling AI)已选定承销行,最早 2027 年上市,募资至少 $1 billion,规模和时间仍可能调整。
信息来源:X:X.PIN (@thexpin) · Kimi
产品发布 / 更新普通
瑞萨推出首款低压氮化镓功率半导体,面向 AI 数据中心与人形机器人
瑞萨电子推出公司首款低压氮化镓功率半导体,为 100V 增强型(E-mode)GaN 分立功率晶体管系列,面向 AI 数据中心、人形机器人、工厂自动化和工业电机驱动等领域。
信息来源:IT之家·人工智能