AI 情报文章归档
浏览 AI圈报 已保存的 28901 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28901
正式分类6
精选内容4413
全部文章
第 154 / 723 页 · 每页 40 条
教程 / 实战普通
The transcript is the record: reading what a coding agent actually did
Coding agents write their own session transcripts to disk. Where Claude Code, Codex and OpenCode keep those files, and how to find the session that belongs to your task.
信息来源:DEV Community · Claude
论文研究普通
Cross-cohort TB classification using clinical data gathered in Uganda and South Africa
We present a first evaluation of machine learning applied to patient clinical and demographic data gathered in two different countries for the purpose of tuberculosis (TB) screening to identify people who would benefit from expensive molec…
信息来源:arXiv
产品发布 / 更新精选
NVIDIA DGX Spark 推出 64GB 版本,10 月 23 日起以 $4,999 开售
NVIDIA 宣布 DGX Spark 推出 64GB 统一内存新配置,10 月 23 日起由 Acer、ASUS、Dell、Gigabyte、HP 和 MSI 发售,起步价 $4,999,支持最高 1000 亿参数模型在端侧运行。
信息来源:NVIDIA Blog(RSS) · NVIDIA
论文研究普通
Learning a Fact Is Not Learning How to Retrieve It
A model trained on "The capital of X is Y" may produce "Y" after "The capital of X is" but fail after "The capital of X:". We call these different ways of eliciting the same fact request forms. To separate learning a fact from retrieving i…
信息来源:arXiv
产品发布 / 更新普通
英伟达发布 64GB 内存版 DGX Spark 桌面 AI 计算机,售价 4999 美元起
英伟达发布 DGX Spark 64GB 内存版桌面 AI 计算机,起售价 4999 美元,10 月 23 日发售,128GB FE 版同步调价至 6950 美元。
信息来源:IT之家·人工智能 · NVIDIA
论文研究普通
UC Berkeley 论文揭示 LLM 上下文更新失败:注意力漂移导致沿用旧值
UC Berkeley 论文《When Context Changes: Understanding Update Failures in LLMs》提出 stale binding 概念,指 LLM 在偏好或截止时间变化后仍沿用旧值,原因是注意力漂移向旧提法。
信息来源:X:Rohan Paul (@rohanpaul_ai)
教程 / 实战普通
I Built an E-Ink Weather Frame for My Kids Who Can't Read Yet
I have three daughters. The oldest is four, the twins are two. None of them can read yet, and every...
信息来源:DEV Community
教程 / 实战普通
AI Made Me Faster. It Also Exposed What I Didn't Understand
There was a point while building Security Lens when I realized I was fixing problems faster than I...
信息来源:DEV Community
行业动态普通
Perplexity 征集开发者平台改进反馈
我们很想听听如何改进我们的开发者平台!
信息来源:X:Aravind Srinivas(Perplexity CEO) (@AravSrinivas)
行业动态普通
上海市委网信办督促平台防范七类涉未成年人心理健康风险,哔哩哔哩、小红书、拼多多等 17 家平台参会
上海市委网信办 9 月 30 日召开未成年人网络保护专题指导会,督促平台重点防范化解七类涉未成年人心理健康风险,包括网络沉迷成瘾、网络欺凌、容貌身材焦虑、自残自杀类有害信息传播、线上社交过度挤占现实交往、AI 换脸与虚拟人设带来的情感操控,以及不良价值观传播。
信息来源:IT之家·人工智能
论文研究普通
Collective Bias Mitigation via Model Routing and Collaboration
Large language models (LLMs) are increasingly deployed in public health, finance, and governance, requiring both accuracy and societal value alignment. Despite recent advances, LLMs often perpetuate or amplify bias embedded in their traini…
信息来源:arXiv
观点 / 方法普通
央视曝光二手平台低价机票陷阱:骗子用真实占座订单行骗,一男子损失近 30 万元
央视财经曝光二手平台低价机票诈骗套路:骗子预先用受害人信息占座下单制造"票已订好"假象,再以订单冻结、需解锁退款为由诱导转账,北京王先生因此损失二十九万余元。警方提醒,陌生人要求下载通联软件或开启屏幕共享的应直接视为诈骗,可拨打 96110 咨询报警。
信息来源:IT之家·人工智能
论文研究普通
WAMpy: Efficient Synthesis of Prolog Programs in Python
We present WAMpy, a Python framework optimized for synthesizing Prolog programs. Unlike general-purpose Prolog systems, WAMpy targets workloads that repeatedly generate and evaluate small candidate programs. WAMpy compiles Prolog clauses i…
信息来源:arXiv
论文研究普通
D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels?
GPU kernels generated by large language model (LLM) agents can remain less efficient than expert implementations, but runtime alone does not reveal how the gap relates to design discovery and implementation. We introduce D2K-Bench, a diagn…
信息来源:arXiv · GPT / Claude / NVIDIA
论文研究普通
The Neuro-Physical Inverter: A Modular Framework for Magnetotelluric Inversion Coupling Ensemble Conditioning with Residual Learning
We present the Neuro-Physical Inverter (NPI), a modular, uncertainty-aware framework for geophysical inversion that couples ensemble-based conditioning with constrained residual learning, demonstrated in the 1D magnetotelluric (MT) setting…
信息来源:arXiv
论文研究普通
Uncertainty as a Proxy for Semantic Correctness in Diffusion-Based Medical Image Synthesis
Diffusion models can synthesise contrast-enhanced CT (CECT) from non-contrast CT (NCCT), avoiding contrast administration and its environmental and patient-access costs. However, visually realistic images are not necessarily anatomically c…
信息来源:arXiv
论文研究普通
AdaStep: Adaptive Step Credit Weighting for Agentic Reinforcement Learning
Long-horizon LLM agents are typically trained with sparse outcome rewards, making trajectory-level objectives too coarse to distinguish the contribution of individual decisions. Step-level credit assignment provides finer-grained supervisi…
信息来源:arXiv
论文研究普通
Near-Optimal Convex Optimization with Lazy Second-Order Oracles
This paper studies the complexity of convex optimization using lazy second-order oracles (Doikov, Chayti, and Jaggi, ICML 2023), where an algorithm queries gradients every iteration and Hessians once per $m$ iterations. Under this setting,…
信息来源:arXiv
论文研究普通
Evolving Hybrid Quantum-Classical Architectures for Image Classification
Hybrid quantum classical neural networks integrate parameterized quantum circuits (PQCs) with established deep learning architectures, but their performance depends strongly on the choice of quantum circuit architecture, a choice that rema…
信息来源:arXiv
教程 / 实战普通
How the GitGuardian Mixin Kit Extends Docker Sandboxes for Safer AI Coding
TL;DR A partnership for security: Our new Sandbox Mixin Kit pairs Docker's isolated...
信息来源:DEV Community
观点 / 方法普通
Krea AI 创意智能体 48 小时构建
@thesystms 团队在 48 小时内构建了一个创意智能体。 @ingi_erlingsson 写下了他如何解读情绪板:先感受,再是色彩、色调和镜头,然后 @armandgw 把它做成了 Krea Agent 技能。 在下方了解详情👇
信息来源:X:Krea AI (@krea_ai)
观点 / 方法普通
推荐一个靠谱的朋友。
推荐一个靠谱的朋友。
信息来源:X:Barret 李靖 (@barret_china)
行业动态普通
湖北智能家居国补扩大范围,机器人、数码相机、无人机、血压仪等可补贴 15%
湖北省对 2026 年消费品以旧换新新增品类补贴政策作出优化调整,自 2026 年 10 月 1 日起实施,落实《商务部等 8 部门关于加快“人工智能 + 消费”发展的实施意见》。
信息来源:IT之家·人工智能
教程 / 实战普通
Building Sarrera: Self-Hosted Enterprise AI Inference Gateway with RBAC, Token Quotas & Telemetry
How to deploy a private, local AI gateway for engineering teams using LiteLLM, Caddy, Langfuse, and Open WebUI with multi-tier quotas.
信息来源:DEV Community
教程 / 实战普通
The n8n Architecture Validator: Resolving API Rate Limit Conflicts with Sanity MCP (Path 1)
For the Sanity Challenge, I decided to tackle Path One: Ship an agent that queries real content. As...
信息来源:DEV Community
观点 / 方法普通
Cohere CEO 批评 Anthropic 游说宗教界强推 AI 立场
Cohere CEO Aidan Gomez 公开批评 Anthropic 游说各大宗教团体采纳其对 AI 的哲学解读,而非听取对方意见,并威胁退出立场不完全一致的活动的做法。他称这是 EA 与 Anthropic 长期以来的道德傲慢与教条主义。引用推文称 Anthropic 曾试图在《Magnifica Humanitas》发布前施压梵蒂冈修改文本,使其不否认 AI 人格。
信息来源:X:Aidan Gomez(Cohere CEO,@aidangomez) · Claude
教程 / 实战普通
GitHub Issues as the Only Database: Running Coding CLIs Unattended Without a Server
ghswarm treats GitHub labels, Issue bodies and checkboxes as the state machine for orchestrating Claude Code / Codex / Cursor headlessly. No webhook, no FastAPI, no database — and it resumes cleanly after the process dies.
信息来源:DEV Community · Claude / Cursor
教程 / 实战普通
Lean Agents: Decide What Your Agent Can Reach Before It Runs
Every tool you connect to an agent is two things at once: tokens the model reads on every run, and an...
信息来源:DEV Community
行业动态普通
三星电子已量产 Exynos 2700 处理器,Galaxy S27 系列将搭载
三星电子已开始量产 Exynos 2700 处理器,产量比上一代提高 10%,该芯片将用于 Galaxy S27 系列。其工程样片 GeekBench 6.7.1 单核 4,328 分、多核 14,700 分,CPU 采用 1+4+1+4 共 10 核设计,最高频率 4.24 GHz,或搭载 Xclipse 970 GPU 与专用 NPU,并支持 LPDDR6 内存。
信息来源:IT之家·人工智能
教程 / 实战普通
Graduating in 2030: Is learning C worth it? Honestly?
I started my degree just 1-2 months ago, which means I will be graduating in 2030. Given how rapidly...
信息来源:DEV Community
产品发布 / 更新普通
SenseNova 6.8 动态设计功能预览
一片静谧的风景,缓缓流动。🌿💧❄️ 借助由 𝗦𝗲𝗻𝘀𝗲𝗡𝗼𝘃𝗮 𝟲.𝟴 𝗙𝗹𝗮𝘀𝗵 𝗣𝗿𝗲𝘃𝗶𝗲𝘄 驱动的 𝗗𝘆𝗻𝗮𝗺𝗶𝗰 𝗗𝗲𝘀𝗶𝗴𝗻,小红书创作者 @Muzlz8888(Mr.木子李子)让沙粒旋舞、溪水流淌、鸭子游动、雪花飘落——同时保持建筑静止不动。 从一张静态图像到一段富有律动的场景。✨
信息来源:X:商汤 SenseTime (@SenseTime_AI)
教程 / 实战普通
I Published a Median Model Lifespan. Seven Weeks Later the Same Query Said 219 Days More.
On 2026-08-01 I computed how long an AI model lives, from the day it shows up to the day its vendor...
信息来源:DEV Community
观点 / 方法普通
AI 幻觉让"顾客永远是对的"变得更糟:服务员、护林员和幼教都在被 ChatGPT 反驳
生成式 AI 的普及正让服务业顾客用 ChatGPT 的答案反驳一线员工:纽约服务员 Madison 称有客人坚称 ChatGPT 说菜品不含贝类,无视其过敏警告;还有客人拒绝侍酒师、改问 ChatGPT,甚至点出不存在的酒。一位护林员表示,多组游客拿着 AI 编造的露营点和逆流路线仍不信地图,幼教也遇到家长坚持 ChatGPT 的睡眠安排。
信息来源:The Verge:AI(RSS) · ChatGPT
行业动态普通
Microsoft 在数据中心扩张仿生态设计,试图缓和社区反弹
Microsoft 宣布把仿生态(biomimicry)方案扩展到美国和德国 20 多个数据中心站点,通过花园、本土树木、传粉花园和植物缓冲带来融入自然景观并降低噪音。
信息来源:The Verge:AI(RSS)
产品发布 / 更新普通
创维 F3 Pro 闺蜜机新增 12GB RAM + 256GB 存储版本,售价 6099 元
创维为 F3 Pro 闺蜜机新增 12GB RAM + 256GB 存储空间版本,定价 6099 元,部分地区国补后低至 4999 元。该机采用 32 英寸 3840x2160 分辨率 IPS LCD 面板,内置 15000mAh 电池、至高 12 小时续航,搭载基于安卓的酷开 AIOS 2.0 系统,支持 AI 健身私教、电子白板、AI 健康智能体等功能。
信息来源:IT之家·人工智能
论文研究普通
NVIDIA 论文测试长程智能体可靠性:128K token 任务准确率比 4K 低 62.8%
NVIDIA 发布论文《Staying on Task: Testing the Foundations of Long-Horizon Agent Reliability》(arxiv.org/abs/2609.38712),提出 Long-Transduction 诊断方法,测试模型在长生成中持续保持任务的能力。
信息来源:X:Rohan Paul (@rohanpaul_ai) · NVIDIA
行业动态普通
AWS CEO 发文警告社区勿阻挠 AI 数据中心建设
AWS CEO Matt Garman 发表超 3000 字博客,呼吁支持 AI 数据中心建设,称失去 AI 主导地位竞争将对美国经济和国家安全造成无法弥补的伤害。文中提到全美有超 100 项数据中心暂停令正在审议,Amazon 承诺投资超 10 亿美元支持数据中心周边社区,并不再与政府机构签署保密协议。
信息来源:The Verge:AI(RSS)
教程 / 实战普通
AcademAI — An Open-Source Research Companion Built for My Wife's PAUD Thesis
Helping students find traceable sources, understand their data, and develop research drafts they can review and defend. Powered by an open-source Node.js agent harness, Google Gemini, and SerpApi.
信息来源:DEV Community · Gemini
产品发布 / 更新普通
Kling 4.0 进入封闭测试,官方发布创意总监商业制作实测
Kling AI 官宣 Kling 4.0 进入封闭测试,计划今年 10 月正式发布,带来更稳定的动态运动、高质量立体声与更准的口型同步,支持最高 4K 分辨率与 10-bit HDR 输出,可生成最长 30 秒视频。
信息来源:X:可灵 Kling AI (@Kling_ai)
教程 / 实战普通
nefuOS: A from-scratch OS with a browser, WebGL, and a local LLM — built in one month
One month ago, I saw someone's hobby OS and thought: "I want to build one too." One month later,...
信息来源:DEV Community