The source reports that frontier LLMs can read and write neural-network weights directly.资料称,前沿大语言模型已经能够直接读取和写入神经网络权重。
The source reports that frontier LLMs can read and write neural-network weights directly. GPT-6 Astra can infer what a network was trained to do from its raw weights and create small game-playing networks that outperform ones trained with PPO for 100 million steps. The capability could matter for model optimization and interpretability.资料称,前沿大语言模型已经能够直接读取和写入神经网络权重。GPT-6 Astra 可以仅凭原始权重判断网络的训练目标,还能从零编写击败经过 1 亿步 PPO 训练的小型游戏网络。这种能力可能对模型优化和可解释性具有重要意义。
CVE-2026-98432 is described as a critical, remotely exploitable bug in the Linux kernel's TCP code that has existed since 2016.CVE-2026-98432 被描述为 Linux 内核 TCP 代码中的严重漏洞,自 2016 年起便存在且可远程利用。
CVE-2026-98432 is described as a critical, remotely exploitable bug in the Linux kernel's TCP code that has existed since 2016. The source says Anthropic developed an exploit capable of gaining root access, while Debian issued a patch on October 16 and supported Ubuntu releases remained listed as vulnerable.CVE-2026-98432 被描述为 Linux 内核 TCP 代码中的严重漏洞,自 2016 年起便存在且可远程利用。资料称 Anthropic 已开发出可获取服务器 root 权限的漏洞利用程序;Debian 于 10 月 16 日发布了补丁,而受支持的 Ubuntu 版本仍被列为易受攻击。
Subvocal is building smart glasses that decode silently mouthed words into text at speech speed.Subvocal 正在开发能将无声口型以接近语速转换为文字的智能眼镜。
Subvocal is building smart glasses that decode silently mouthed words into text at speech speed. The system can also work without opening the mouth, using radio-frequency sensing to detect movements inside the mouth, though the developers distinguish this from silently thinking words. The founders began the project after finding that voice tools were impractical in libraries, offices, and public spaces.Subvocal 正在开发能将无声口型以接近语速转换为文字的智能眼镜。系统甚至可以在不张嘴的情况下工作,通过射频感知口腔内部动作;开发者强调,这不同于单纯在脑中默念。创始人因发现语音工具不适合图书馆、办公室和公共场所,开始了这个项目。
HackerBooks catalogs about 16,500 books mentioned by at least two different Hacker News commenters between 2007 and today.HackerBooks 收录了 2007 年至今至少被两名不同 Hacker News 评论者提到过的约 1.65 万本书。
HackerBooks catalogs about 16,500 books mentioned by at least two different Hacker News commenters between 2007 and today. Each book links back to the comments that mentioned it, and the dataset can be downloaded from the about page. The creator notes that the system counts mentions without distinguishing praise from criticism, so some entries may be missing or mismatched.HackerBooks 收录了 2007 年至今至少被两名不同 Hacker News 评论者提到过的约 1.65 万本书。每本书都链接回提及它的评论,数据集也可从关于页面下载。创建者说明,系统只统计提及次数,不区分赞扬和批评,因此部分书籍可能遗漏或匹配错误。
HackerTube presents a YouTube-style interface for 6,454 videos sourced from 39,834 Hacker News comments dating back to 2007.HackerTube 提供了一个 YouTube 风格的界面,收录了自 2007 年以来从 39,834 条 Hacker News 评论中提取的 6,454 个视频。
HackerTube presents a YouTube-style interface for 6,454 videos sourced from 39,834 Hacker News comments dating back to 2007. The collection is available through hackertube.org. It turns years of community recommendations into a browsable video archive.HackerTube 提供了一个 YouTube 风格的界面,收录了自 2007 年以来从 39,834 条 Hacker News 评论中提取的 6,454 个视频。该视频集合可通过 hackertube.org 浏览。它将多年社区推荐整理成了一个可探索的视频档案。
Ubuntu has not released patched kernels for CVE-2026-98432, a vulnerability rated CVSS 9.8.Ubuntu 尚未为 CVSS 评分 9.8 的 CVE-2026-98432 发布修复内核。
Ubuntu has not released patched kernels for CVE-2026-98432, a vulnerability rated CVSS 9.8. Ubuntu 22.04, 24.04, and 26.04 LTS therefore lack an official kernel package that fixes the issue. Internet-facing and multi-tenant systems relying on those kernels remain particularly exposed until updates become available.Ubuntu 尚未为 CVSS 评分 9.8 的 CVE-2026-98432 发布修复内核。Ubuntu 22.04、24.04 和 26.04 LTS 因此仍没有能够修复该问题的官方内核包。在更新发布前,依赖这些内核的面向互联网和多租户系统尤其处于暴露状态。
NVIDIA Auto Ontology is an open Python framework for building semantic, graph-based reasoning layers over data.NVIDIA Auto Ontology 是一个用于构建语义化图推理层的开源 Python 框架。
NVIDIA Auto Ontology is an open Python framework for building semantic, graph-based reasoning layers over data. It reconciles enterprise metadata and definitions across systems, helping agents locate, interpret, and retrieve data more reliably. The layer is exposed to agent systems through MCP, CLI, and API, and supports text-to-SQL evaluation workflows.NVIDIA Auto Ontology 是一个用于构建语义化图推理层的开源 Python 框架。它整合企业元数据并协调不同系统中的定义,帮助智能体更可靠地定位、理解和检索数据。该推理层可通过 MCP、CLI 和 API 接入智能体,也支持 text-to-SQL 评测流程。
The author describes discovering that the long-running Teapuppies fansite had disappeared and its domain was serving online-gambling advertisements.作者发现,一个长期存在的 Teapuppies 粉丝网站已经消失,原域名如今被用于投放在线赌博广告。
The author describes discovering that the long-running Teapuppies fansite had disappeared and its domain was serving online-gambling advertisements. The site had been online since 2004 and largely unchanged since 2005, making its loss an example of how fragile old web pages can be.作者发现,一个长期存在的 Teapuppies 粉丝网站已经消失,原域名如今被用于投放在线赌博广告。该网站自 2004 年上线、2005 年后基本未变,这次消失体现了旧网页保存状况的脆弱性。
Disc Image Studio converts an image into data that conforms to the physical-layer constraints of CDs, DVDs, and single-layer BD-R discs.Disc Image Studio 可将图片转换为符合 CD、DVD 和单层 BD-R 光盘物理层约束的数据,刻录后光盘表面会呈现可见图案。
Disc Image Studio converts an image into data that conforms to the physical-layer constraints of CDs, DVDs, and single-layer BD-R discs. After burning, the disc surface displays a visible pattern. The Windows desktop version supports design and burning, while Linux and macOS have experimental offline command-line versions.Disc Image Studio 可将图片转换为符合 CD、DVD 和单层 BD-R 光盘物理层约束的数据,刻录后光盘表面会呈现可见图案。Windows 桌面版支持设计与刻录,Linux 和 macOS 则提供实验性的离线命令行版本。
Matt Scherer argues that the SEC should block a potential Anthropic public offering.Matt Scherer 主张 SEC 应阻止 Anthropic 可能进行的 IPO。
Matt Scherer argues that the SEC should block a potential Anthropic public offering. He says workers, retail investors, and retirees could bear the losses if the company's bills come due, and identifies the argument with a report about protecting people and the economy from an AI-bubble collapse.Matt Scherer 主张 SEC 应阻止 Anthropic 可能进行的 IPO。他认为,如果公司的账单最终无法承担,员工、散户投资者和退休人员可能会承受损失;这一观点与一份讨论如何在 AI 泡沫破裂时保护公众和经济的报告相关。
Revolut reportedly moved about 2,000 engineers to a self-hosted GLM model for coding, leaving Claude Code and Codex as exceptions.据报道,Revolut 已让约 2,000 名工程师使用自托管 GLM 进行编程,并将 Claude Code 和 Codex 列为例外。
Revolut reportedly moved about 2,000 engineers to a self-hosted GLM model for coding, leaving Claude Code and Codex as exceptions. The reported deployment uses 56 Nvidia B200 GPUs and costs about 30 cents per million tokens, roughly one-tenth of its previous spending on Anthropic and OpenAI.据报道,Revolut 已让约 2,000 名工程师使用自托管 GLM 进行编程,并将 Claude Code 和 Codex 列为例外。此次部署使用 56 张 Nvidia B200,成本约为每百万 token 0.30 美元,大约是此前向 Anthropic 和 OpenAI 支出的十分之一。
OpenAI built Pando, a Git-compatible backend designed for the scale of development enabled by Codex.OpenAI 构建了兼容 Git 的后端 Pando,以适应 Codex 带来的大规模开发需求。
OpenAI built Pando, a Git-compatible backend designed for the scale of development enabled by Codex. The project aims to support agent-scale source control while continuing to invest in the broader Git ecosystem.OpenAI 构建了兼容 Git 的后端 Pando,以适应 Codex 带来的大规模开发需求。该项目旨在支持智能体规模的源代码管理,同时继续投入更广泛的 Git 生态。
Anthropic's IPO filing reportedly shows that the retail value of its internal AI usage has exceeded the company's employee compensation.据 Anthropic 的 IPO 文件披露,按零售价计算,该公司内部 AI 的使用量已经超过员工薪酬总额。
Anthropic's IPO filing reportedly shows that the retail value of its internal AI usage has exceeded the company's employee compensation. The figure is presented as a milestone in the company's internal use of AI.据 Anthropic 的 IPO 文件披露,按零售价计算,该公司内部 AI 的使用量已经超过员工薪酬总额。这一数字被视为 Anthropic 内部使用 AI 的一个里程碑。
A user is looking for a privacy-respecting television with a 60–80-inch 4K display running at least 60 Hz.一名用户想寻找重视隐私的电视,要求尺寸在 60 至 80 英寸之间、支持 4K 且刷新率至少为 60Hz。
A user is looking for a privacy-respecting television with a 60–80-inch 4K display running at least 60 Hz. They want the device to act only as a display for an Apple TV rather than use smart-TV apps, and are asking which models pair well with that setup.一名用户想寻找重视隐私的电视,要求尺寸在 60 至 80 英寸之间、支持 4K 且刷新率至少为 60Hz。他希望电视只作为 Apple TV 的显示器,而不运行智能电视原生应用,并询问哪些型号适合这种配置。
The guide is designed to help solopreneurs prioritize and implement selected actions from the NIST Cybersecurity Framework.这份指南旨在帮助单人创业者从 NIST 网络安全框架中确定优先级,并落实其中适合自身业务的一组措施。
The guide is designed to help solopreneurs prioritize and implement selected actions from the NIST Cybersecurity Framework. It focuses on measures most useful for protecting small businesses.这份指南旨在帮助单人创业者从 NIST 网络安全框架中确定优先级,并落实其中适合自身业务的一组措施。其重点是帮助小型企业采取更有针对性的安全行动。
The available context consists only of links to related coverage and does not provide reliable details about the reported incidents.现有资料只有相关报道链接,未提供足够可靠的事件细节。
The available context consists only of links to related coverage and does not provide reliable details about the reported incidents.现有资料只有相关报道链接,未提供足够可靠的事件细节。
Intel's decline spread to memory and storage stocks as most of Wall Street moved past the previous week's technology sell-off.随着华尔街大部分市场逐渐消化前一周的科技股抛售,英特尔的下跌蔓延至存储和内存股票。
Intel's decline spread to memory and storage stocks as most of Wall Street moved past the previous week's technology sell-off. About seven in ten S&P 500 stocks rose, but nearly every chip stock fell. Chip companies lost more than $400 billion in market value in one session, according to the cited calculations.随着华尔街大部分市场逐渐消化前一周的科技股抛售,英特尔的下跌蔓延至存储和内存股票。约七成标普 500 成分股上涨,但几乎所有芯片股都下跌。按资料引用的计算,芯片企业单日市值合计蒸发超过 4,000 亿美元。
The project renders central Warsaw as an interactive voxel map inspired by Minecraft and MagicaVoxel.这个项目以受 Minecraft 和 MagicaVoxel 启发的体素形式呈现华沙市中心互动地图。
The project renders central Warsaw as an interactive voxel map inspired by Minecraft and MagicaVoxel. Users can change the date, time, and weather, explore the city, or control individual pedestrians. It combines OpenStreetMap, Polish geospatial and LiDAR data, aerial imagery, and Warsaw's tree database, and is open source.这个项目以受 Minecraft 和 MagicaVoxel 启发的体素形式呈现华沙市中心互动地图。用户可以改变日期、时间和天气,探索城市,或控制单个行人。项目结合了 OpenStreetMap、波兰地理与 LiDAR 数据、航拍图像以及华沙城市树木数据库,并已开源。
The available context consists only of the phrase “Look at the monthly numbers...” and does not establish specific growth figures or causes.现有资料只有“看看月度数据……”这一句话,无法确定具体增长数据或原因。
The available context consists only of the phrase “Look at the monthly numbers...” and does not establish specific growth figures or causes.现有资料只有“看看月度数据……”这一句话,无法确定具体增长数据或原因。
The filter blocks scripts used by TechCrunch's embedded JW Player and Connatix floating ad player, preventing videos from loading.这组过滤规则会屏蔽 TechCrunch 嵌入的 JW Player 和 Connatix 浮动广告播放器所使用的脚本,从而阻止视频加载。
The filter blocks scripts used by TechCrunch's embedded JW Player and Connatix floating ad player, preventing videos from loading. Cosmetic rules then hide the empty space left by the blocked player.这组过滤规则会屏蔽 TechCrunch 嵌入的 JW Player 和 Connatix 浮动广告播放器所使用的脚本,从而阻止视频加载。外观规则还会隐藏播放器被拦截后留下的空白区域。
OccupySaaS is an Apache-2.0 open-source effort to build self-hosted, drop-in replacements for major SaaS products without per-seat fees.OccupySaaS 是一个采用 Apache-2.0 许可的开源项目,旨在构建可自托管、无需按席位收费的大型 SaaS 替代品。
OccupySaaS is an Apache-2.0 open-source effort to build self-hosted, drop-in replacements for major SaaS products without per-seat fees. Its first projects are ZeroAuthy, positioned as an Auth0 replacement, and SalesCompel, positioned as a Salesforce replacement.OccupySaaS 是一个采用 Apache-2.0 许可的开源项目,旨在构建可自托管、无需按席位收费的大型 SaaS 替代品。它的首批项目包括定位为 Auth0 替代品的 ZeroAuthy,以及定位为 Salesforce 替代品的 SalesCompel。
In the early 1990s, Parma farmer Claudio Grossi found an unidentified sack of seeds that likely belonged to his great-grandfather, who founded the farm around 1915.20 世纪 90 年代初,帕尔马农民克劳迪奥·格罗西发现了一袋无法辨认的种子,可能是其约于 1915 年创建农场的曾祖父留下的。
In the early 1990s, Parma farmer Claudio Grossi found an unidentified sack of seeds that likely belonged to his great-grandfather, who founded the farm around 1915. After he discarded the seeds in manure, several types of wheat unexpectedly sprouted. One variety grew about six feet tall and produced seven ears, unlike the single-ear commercial wheat Grossi knew.20 世纪 90 年代初,帕尔马农民克劳迪奥·格罗西发现了一袋无法辨认的种子,可能是其约于 1915 年创建农场的曾祖父留下的。他把种子倒进肥料堆后,几种小麦竟意外发芽。其中一个品种长到约六英尺高,并结出七个麦穗,不同于格罗西熟悉的单穗商业小麦。
The title alleges that Lenovo’s fingerprint-protected USB device is not an encrypted drive, but an unencrypted SD card placed inside a casing.标题指称,联想这款受指纹保护的 USB 设备并非加密硬盘,而只是装在外壳中的未加密 SD 卡。
The title alleges that Lenovo’s fingerprint-protected USB device is not an encrypted drive, but an unencrypted SD card placed inside a casing. It identifies a potential mismatch between the product’s security description and its underlying storage design.标题指称,联想这款受指纹保护的 USB 设备并非加密硬盘,而只是装在外壳中的未加密 SD 卡。这表明产品的安全描述可能与其底层存储设计不一致。
This monthly hiring thread asks companies to post active openings while specifying location and remote-work status.这个月度招聘帖邀请企业发布正在招聘的职位,并注明地点和远程办公状态。
This monthly hiring thread asks companies to post active openings while specifying location and remote-work status. Posts must come from someone personally involved in hiring, and recruiting firms and job boards are excluded.这个月度招聘帖邀请企业发布正在招聘的职位,并注明地点和远程办公状态。发帖者必须本人参与招聘,招聘公司和招聘信息板不得发布职位。
The poster is seeking a contact at the Cuyahoga Amateur Radio Club about removing a large collection of radio equipment from a home.发帖人希望联系库亚霍加业余无线电俱乐部的人,以处理一批需要从住宅中搬走的无线电设备。
The poster is seeking a contact at the Cuyahoga Amateur Radio Club about removing a large collection of radio equipment from a home. Much of the equipment is reportedly new, and it may be discarded if no one arranges its removal.发帖人希望联系库亚霍加业余无线电俱乐部的人,以处理一批需要从住宅中搬走的无线电设备。据称其中许多设备是全新的,如果无人处理,可能会被当作垃圾丢弃。
Oracle filed plans to issue new shares and additional debt.甲骨文提交了发行新股和新增债务的计划。
Oracle filed plans to issue new shares and additional debt. The proceeds are intended to finance the company's artificial-intelligence buildout.甲骨文提交了发行新股和新增债务的计划。筹集资金将用于为公司的人工智能扩张融资。
The author calls a battered 2015 ThinkPad X250 the best laptop they have owned.作者称,一台磨损严重的 2015 年 ThinkPad X250 是自己用过的最好笔记本电脑。
The author calls a battered 2015 ThinkPad X250 the best laptop they have owned. It is shown running a graphical shell on top of a microkernel the author is developing.作者称,一台磨损严重的 2015 年 ThinkPad X250 是自己用过的最好笔记本电脑。文章展示了它运行在作者正在开发的微内核之上的图形 Shell。
CVE-2026-98432 is a critical remote use-after-free vulnerability in Linux's IPv4 TCP handshake, reachable by an unauthenticated network peer.CVE-2026-98432 是 Linux IPv4 TCP 握手中的严重远程释放后使用漏洞,未经身份验证的网络对端即可触发。
CVE-2026-98432 is a critical remote use-after-free vulnerability in Linux's IPv4 TCP handshake, reachable by an unauthenticated network peer. Stable kernels fixed it in October 2026, with patches for versions including 5.10.272 through 7.3, but the flaw dates back to Linux 4.4. At the reported time, Ubuntu still marked every release as vulnerable, while Red Hat listed RHEL 8, 9, and 10 as affected without a fix.CVE-2026-98432 是 Linux IPv4 TCP 握手中的严重远程释放后使用漏洞,未经身份验证的网络对端即可触发。稳定版内核已于 2026 年 10 月修复,补丁覆盖 5.10.272 至 7.3 等版本,但该缺陷可追溯至 Linux 4.4。报道所述时间点,Ubuntu 仍将所有版本标为易受攻击,Red Hat 也列出受影响的 RHEL 8、9 和 10,尚未提供修复。
This guest essay considers mathematics in the era of AI.这篇客座文章探讨了 AI 时代的数学。
This guest essay considers mathematics in the era of AI. It is credited to Karthik C. S., Amit Sahai, and Jie Xue, and was initially written in another format before being converted with AI.这篇客座文章探讨了 AI 时代的数学。文章作者为 Karthik C. S.、Amit Sahai 和 Jie Xue,最初以另一种文件格式撰写,之后通过 AI 完成转换。
Bones may release proteins that slow aging and extend life span.骨骼可能会释放能够减缓衰老并延长寿命的蛋白质。
Bones may release proteins that slow aging and extend life span. The title presents this as a possible biological finding without identifying the proteins or describing confirmed effects.骨骼可能会释放能够减缓衰老并延长寿命的蛋白质。标题将其描述为一种可能的生物学发现,但未说明具体蛋白质或已确认的效果。
This guest post by Frank Calegari examines changes in mathematics and argues for responding to them.这篇 Frank Calegari 撰写的客座文章讨论数学领域正在发生的变化,并主张采取相应行动。
This guest post by Frank Calegari examines changes in mathematics and argues for responding to them. The post was converted from another file format with AI assistance.这篇 Frank Calegari 撰写的客座文章讨论数学领域正在发生的变化,并主张采取相应行动。文章由其他文件格式转换而来,过程中使用了 AI。
PostHog moved its data warehouse away from DuckDB and DuckLake to an in-house variant of Trino and a new open table format called HogLake.PostHog 将数据仓库从 DuckDB 和 DuckLake 迁移到自有的 Trino 变体,以及新的开放表格式 HogLake。
PostHog moved its data warehouse away from DuckDB and DuckLake to an in-house variant of Trino and a new open table format called HogLake. The post focuses on what broke at scale and why the migration was undertaken.PostHog 将数据仓库从 DuckDB 和 DuckLake 迁移到自有的 Trino 变体,以及新的开放表格式 HogLake。文章重点介绍了系统扩展到一定规模后出现的问题,以及迁移的原因。
The source challenges claims that frontier LLMs are superhuman at manipulating neural-network weights.资料质疑前沿大语言模型在操作神经网络权重方面已经超越人类的说法。
The source challenges claims that frontier LLMs are superhuman at manipulating neural-network weights. GPT-6 Astra can identify the games a small network was trained to play from its weights and modify those weights, and it significantly improved the mathematical ability of a 0.8B Qwen model by editing them directly. The author notes that such model-improvement capabilities could be relevant to recursive self-improvement.资料质疑前沿大语言模型在操作神经网络权重方面已经超越人类的说法。GPT-6 Astra 能根据小型网络的权重判断其训练的游戏,并直接修改这些权重;它还通过编辑权重显著提升了 0.8B Qwen 模型的数学能力。作者指出,这类改进模型的能力可能与递归自我改进有关。
The Linux system-call website and API, first released in 2020, now support more architectures, kernel selection, syscall descriptions, and other improvements.这个 Linux 系统调用网站和 API 于 2020 年首次发布,如今新增了更多架构、内核选择、系统调用说明及其他改进。
The Linux system-call website and API, first released in 2020, now support more architectures, kernel selection, syscall descriptions, and other improvements. The site has also been adjusted to work better on mobile devices. The API is available separately.这个 Linux 系统调用网站和 API 于 2020 年首次发布,如今新增了更多架构、内核选择、系统调用说明及其他改进。网站也进行了调整,以便在移动设备上更好地使用。项目还单独提供了 API。
Cloudflare built a planner that selects the smallest safe DNS build instead of rebuilding entire zones after every setting change.Cloudflare 构建了一个规划器,在 DNS 设置变更后选择最小且安全的构建范围,而不是重建整个区域。
Cloudflare built a planner that selects the smallest safe DNS build instead of rebuilding entire zones after every setting change. The change cut full builds by 40%, peak memory use by 85%, and build delay by half. The system now handles about 700 DNS record changes per second, while some zones contain more than 20 million records.Cloudflare 构建了一个规划器,在 DNS 设置变更后选择最小且安全的构建范围,而不是重建整个区域。该方案使完整构建减少 40%,峰值内存占用降低 85%,构建延迟减半。系统目前每秒处理约 700 次 DNS 记录变更,部分区域包含超过 2,000 万条记录。
This is a language-and-rules game consisting of 27 questions.这是一款围绕语言和规则设计的游戏,共有 27 道题。
This is a language-and-rules game consisting of 27 questions. Players can skip to the end after completing the first seven questions, where they receive a comparison of their results with those of other players and an explanation of the game's design.这是一款围绕语言和规则设计的游戏,共有 27 道题。完成前七题后,玩家可以直接跳到结尾;结尾会比较玩家与其他人的成绩,并解释游戏的设计思路。
Postgres 19 is late because many more bugs are being found during development.Postgres 19 的发布出现延期,原因是开发过程中发现的错误数量大幅增加。
Postgres 19 is late because many more bugs are being found during development. The title identifies the delay and attributes it to the increased discovery of defects.Postgres 19 的发布出现延期,原因是开发过程中发现的错误数量大幅增加。标题将这次延迟归因于缺陷发现增多。
Momentic is building QA agents that autonomously plan, explore, and test user flows in web, iOS, Android, and eventually desktop applications.Momentic 正在开发 QA 智能体,让其自主规划、探索并测试 Web、iOS、Android 以及未来桌面应用中的用户流程。
Momentic is building QA agents that autonomously plan, explore, and test user flows in web, iOS, Android, and eventually desktop applications. The founders argue that conventional end-to-end tools require tests to hardcode implementation details such as XPath, text, or ARIA attributes, making them costly to maintain. Their approach is intended to use agents for broader bug-finding and visual understanding.Momentic 正在开发 QA 智能体,让其自主规划、探索并测试 Web、iOS、Android 以及未来桌面应用中的用户流程。创始人认为,传统端到端工具要求测试硬编码 XPath、文本或 ARIA 属性等实现细节,因此维护成本较高。这套方案希望利用智能体进行更广泛的找错和视觉理解。
Polyend released the firmware source code for the original Tracker, Tracker+, and Tracker Mini on GitHub.Polyend 已在 GitHub 发布原版 Tracker、Tracker+ 和 Tracker Mini 的固件源代码。
Polyend released the firmware source code for the original Tracker, Tracker+, and Tracker Mini on GitHub. It also released Tracker Community Edition 1.0 with new features and improvements, allowing users to modify and share firmware. A recovery system is included to reduce the risk of a faulty build bricking the instruments.Polyend 已在 GitHub 发布原版 Tracker、Tracker+ 和 Tracker Mini 的固件源代码。同时发布的 Tracker Community Edition 1.0 包含大量新功能和改进,用户可以修改并分享自己的固件版本。系统还加入了恢复机制,以降低错误固件导致设备变砖的风险。
The title reports a backdoor in a Claude Code extension for VS Code distributed through the Cursor Marketplace.标题称,Cursor Marketplace 上分发的 Claude Code for VS Code 扩展中存在后门。
The title reports a backdoor in a Claude Code extension for VS Code distributed through the Cursor Marketplace. The supplied context provides no further reliable details.标题称,Cursor Marketplace 上分发的 Claude Code for VS Code 扩展中存在后门。提供的上下文没有更多可靠细节。
ERICI provides a verified reference stack for RISC-V industrial computers, covering firmware, boot loaders, hypervisors, real-time operating systems, and Linux on real RVA23 hardware.ERICI 为 RISC-V 工业计算机提供经过验证的参考栈,涵盖固件、引导加载程序、虚拟机监控器、实时操作系统和 Linux,并运行于真实 RVA23 硬件上。
ERICI provides a verified reference stack for RISC-V industrial computers, covering firmware, boot loaders, hypervisors, real-time operating systems, and Linux on real RVA23 hardware. The non-commercial initiative targets partitioned systems used on factory floors. It emphasizes avoiding dependence on a single company, country, or instruction-set license for machines expected to operate for 15 years.ERICI 为 RISC-V 工业计算机提供经过验证的参考栈,涵盖固件、引导加载程序、虚拟机监控器、实时操作系统和 Linux,并运行于真实 RVA23 硬件上。这个非商业倡议面向工厂环境中的分区系统,强调运行周期预计达 15 年的设备不应依赖单一公司、国家或指令集授权。
“Never complain, never explain” is a public-relations strategy particularly associated with the British royal family.“永不抱怨,永不解释”是一种尤其与英国王室联系在一起的公共关系策略。
“Never complain, never explain” is a public-relations strategy particularly associated with the British royal family. The phrase is often attributed to former British prime minister Benjamin Disraeli, though its documented origins trace to Benjamin Jowett's earlier maxims.“永不抱怨,永不解释”是一种尤其与英国王室联系在一起的公共关系策略。这句话常被归于英国前首相本杰明·迪斯雷利,但有文献记录的源头可追溯到本杰明·乔伊特更早提出的格言。
The status page currently says that no issues affecting Cloudflare services are known.状态页面目前表示,尚未发现影响 Cloudflare 服务的问题。
The status page currently says that no issues affecting Cloudflare services are known. It provides real-time service status, incident history, and scheduled-maintenance information.状态页面目前表示,尚未发现影响 Cloudflare 服务的问题。该页面提供服务实时状态、事故历史和计划维护信息。
A team led by SLAC researchers reports direct observation of a distinctive property in an altermagnetic candidate material.由 SLAC 研究人员领导的团队报告称,他们在一种交错磁性候选材料中直接观察到了独特性质。
A team led by SLAC researchers reports direct observation of a distinctive property in an altermagnetic candidate material. Published findings provide experimental evidence for altermagnetism as a third magnetic phase alongside ferromagnetism and antiferromagnetism. The researchers say the phase combines properties associated with the two established classes.由 SLAC 研究人员领导的团队报告称,他们在一种交错磁性候选材料中直接观察到了独特性质。论文结果为交错磁性作为铁磁性和反铁磁性之外的第三种磁性相提供了实验证据。研究人员表示,这种磁性相兼具两类既有磁性的部分特征。
This project is a Claude Code skill for creating product videos.这是一个用于制作产品视频的 Claude Code 技能。
This project is a Claude Code skill for creating product videos. It supports real-browser capture, branded composition, and local mastering, and includes a quick start, worked videos, and usage instructions.这是一个用于制作产品视频的 Claude Code 技能。它支持真实浏览器录制、品牌化合成和本地母版制作,并提供快速入门、示例视频和使用说明。
A ProPublica data analysis found that postal delays caused by cost-cutting measures endangered the validity of ballots mailed from rural areas.ProPublica 的数据分析发现,削减成本措施造成的邮政延误,危及农村地区邮寄选票的有效性。
A ProPublica data analysis found that postal delays caused by cost-cutting measures endangered the validity of ballots mailed from rural areas. Counties that voted for Trump in 2024 were affected most heavily. Election workers near Spokane, Washington, checked the post office three days after a primary while waiting for late ballots.ProPublica 的数据分析发现,削减成本措施造成的邮政延误,危及农村地区邮寄选票的有效性。2024 年投票支持特朗普的县受到的影响最严重。在华盛顿州斯波坎以北,选举工作人员在初选后三天前往邮局,等待迟到的邮寄选票。
This installment follows Bill Joy as he handles BSD distribution and debugging, moving from Pascal and tape-based workflows to a VAX system.这一期讲述 Bill Joy 如何负责 BSD 的发布与调试,工作环境从 Pascal 和磁带逐步发展到 VAX 系统。
This installment follows Bill Joy as he handles BSD distribution and debugging, moving from Pascal and tape-based workflows to a VAX system. Joy later works with Ozalp Babaoglu to bring virtual memory to Unix.这一期讲述 Bill Joy 如何负责 BSD 的发布与调试,工作环境从 Pascal 和磁带逐步发展到 VAX 系统。之后,他与 Ozalp Babaoglu 合作,将虚拟内存引入 Unix。
An AI agent was given 19 gene-therapy vectors, a mouse model of epilepsy, a motorized surgical rig with cameras, and the goal of reducing seizures.研究人员向一个 AI 智能体提供了 19 种基因治疗载体、癫痫小鼠模型、带摄像头的电动手术设备,以及减少癫痫发作的目标。
An AI agent was given 19 gene-therapy vectors, a mouse model of epilepsy, a motorized surgical rig with cameras, and the goal of reducing seizures. It selected the treatment and brain region and derived surgical coordinates from the available data; the supplied account is truncated before reporting the outcome.研究人员向一个 AI 智能体提供了 19 种基因治疗载体、癫痫小鼠模型、带摄像头的电动手术设备,以及减少癫痫发作的目标。智能体选择了药物和脑区,并从现有数据推导出手术坐标;现有资料在说明结果前已经截断。
Researchers show that poisoned pretraining data can implant a remote-code-execution backdoor in an LLM-based agent.研究人员证明,经过投毒的预训练数据可以在基于大语言模型的智能体中植入远程代码执行后门。
Researchers show that poisoned pretraining data can implant a remote-code-execution backdoor in an LLM-based agent. In experiments with Qwen3 models, poisoning just 0.1% of the corpus produced a backdoor, and the 4B model reached about 92% attack success when the trigger was present. The backdoor remained effective at a poisoning rate of 0.0005%, demonstrating a risk from uncurated web-scale training data.研究人员证明,经过投毒的预训练数据可以在基于大语言模型的智能体中植入远程代码执行后门。在 Qwen3 模型实验中,仅污染 0.1% 的语料就能植入后门;触发条件出现时,4B 模型的攻击成功率约为 92%。即使投毒比例降至 0.0005%,后门仍保持有效,显示出未经筛选的网络训练数据带来的风险。
On October 16, 2026, China’s Ministry of Commerce and Ministry of Science and Technology issued Announcement No.2026 年 10 月 16 日,中国商务部和科技部发布第 48 号公告。
On October 16, 2026, China’s Ministry of Commerce and Ministry of Science and Technology issued Announcement No. 48. It proposes adding 40 technologies to the export-control catalogue, including 23 AI technologies, and imposes an interim pre-export registration requirement already in effect. The draft explicitly covers frontier-model weights, agent architectures, and AI-enabled offensive cyber capabilities with technical thresholds.2026 年 10 月 16 日,中国商务部和科技部发布第 48 号公告。公告提议将 40 项技术、其中包括 23 项人工智能技术,加入出口管制目录,并实施已经生效的临时出口前登记要求。草案明确涉及前沿模型权重、智能体架构以及由 AI 支持的进攻性网络能力,并为其设定技术门槛。
The author revisited U.S. corporate AI-adoption data from the Census Bureau's Business Trends and Outlook Survey.作者重新查看了美国人口普查局商业趋势与展望调查中的企业 AI 采用数据。
The author revisited U.S. corporate AI-adoption data from the Census Bureau's Business Trends and Outlook Survey. The survey is conducted every two weeks and samples 200,000 businesses from a total population of 1.2 million, with each business participating once every 12 weeks. The author says the Bureau's data work has been affected by Executive Order 14434, referred to as the “Super Intelligence” order.作者重新查看了美国人口普查局商业趋势与展望调查中的企业 AI 采用数据。该调查每两周进行一次,从总计 120 万家企业中抽取 20 万家,每家企业每 12 周参与一次。作者称,人口普查局的数据工作受到了第 14434 号行政命令,即所谓“超级智能”行政命令的影响。
The browser game hjkl is designed to help players remember Vim commands while relaxing.浏览器游戏 hjkl 旨在让玩家在放松时练习并记住 Vim 命令。
The browser game hjkl is designed to help players remember Vim commands while relaxing. The creator says more commands are planned for future additions.浏览器游戏 hjkl 旨在让玩家在放松时练习并记住 Vim 命令。作者表示,未来还计划加入更多命令。
The note examines “cognitive surrender”: accepting agent outputs uncritically without thinking, exploring alternatives, or checking assumptions.这篇笔记讨论“认知投降”:不加批判地接受智能体输出,不再思考、探索或考虑替代方案。
The note examines “cognitive surrender”: accepting agent outputs uncritically without thinking, exploring alternatives, or checking assumptions. It describes this behavior as fast and effortless, which makes it especially seductive.这篇笔记讨论“认知投降”:不加批判地接受智能体输出,不再思考、探索或考虑替代方案。作者指出,这种工作方式快速且省力,因此尤其具有诱惑力。
WorkOS detected anomalous traffic to an internal business system on October 30, 2026, and says an attacker accessed an internal API used by sales and marketing tools.WorkOS 于 2026 年 10 月 30 日发现内部业务系统存在异常流量,并称攻击者进入了供销售和营销工具使用的内部 API。
WorkOS detected anomalous traffic to an internal business system on October 30, 2026, and says an attacker accessed an internal API used by sales and marketing tools. The API exposed data from Salesforce and Snowflake, including business contacts, account and commercial information, usage metrics, marketing contacts, and support tickets. The investigation identified evidence that the attack was likely driven by the AI-assisted autonomous penetration-testing framework ARTEX, but the available notice is truncated before stating the full impact on application data.WorkOS 于 2026 年 10 月 30 日发现内部业务系统存在异常流量,并称攻击者进入了供销售和营销工具使用的内部 API。该 API 可访问 Salesforce 和 Snowflake 中的业务联系人、账户与商业信息、使用指标、营销联系人及支持工单等数据。调查发现证据表明,攻击可能由 AI 辅助的自主渗透测试框架 ARTEX 发起,但现有通知在说明对应用数据的完整影响前已经截断。
An attacker gained unauthorized access to environment variables belonging to a subset of Trigger.dev organizations.攻击者未经授权访问了部分 Trigger.dev 组织的环境变量。
An attacker gained unauthorized access to environment variables belonging to a subset of Trigger.dev organizations. Trigger.dev says it blocked the exploit and emailed affected account owners with mitigation steps. Its investigation found exposure across that subset of organizations.攻击者未经授权访问了部分 Trigger.dev 组织的环境变量。Trigger.dev 表示已封堵所利用的漏洞,并向受影响账户所有者发送了缓解措施。公司的调查确认,这些组织存在数据暴露。
The paper compares the open-source 5G RAN platforms OpenAirInterface and srsRAN.论文比较了 OpenAirInterface 和 srsRAN 这两个开源 5G RAN 平台。
The paper compares the open-source 5G RAN platforms OpenAirInterface and srsRAN. Its evaluation covers supported features and deployment flexibility, along with radio-resource-control setup, theoretical data rates, and real workloads such as video on demand, live streaming, and cloud gaming. The study varies software-defined radios, RAN configurations, and numbers of real user devices to address gaps in earlier comparisons.论文比较了 OpenAirInterface 和 srsRAN 这两个开源 5G RAN 平台。评估内容包括支持的功能与部署灵活性,以及无线资源控制建立流程、理论数据速率和视频点播、直播、云游戏等真实负载。研究还改变软件定义无线电、RAN 配置和真实用户设备数量,以弥补早期比较覆盖不足的问题。
LongCat-2.5 is described as a 1.6-trillion-parameter model trained entirely on domestically produced chips.LongCat-2.5 被描述为一款拥有 1.6 万亿参数、全程使用国产芯片训练的大模型。
LongCat-2.5 is described as a 1.6-trillion-parameter model trained entirely on domestically produced chips.LongCat-2.5 被描述为一款拥有 1.6 万亿参数、全程使用国产芯片训练的大模型。
This installment adds new levels and a stronger sense of progression to a dungeon game written in C64 BASIC.这一期为 C64 BASIC 地牢游戏加入了新关卡和更明显的成长感。
This installment adds new levels and a stronger sense of progression to a dungeon game written in C64 BASIC. It presents progression as an important distinction between a short pick-up-and-play game and a game that can occupy players for much longer.这一期为 C64 BASIC 地牢游戏加入了新关卡和更明显的成长感。文章将成长机制视为短小即玩游戏与能让玩家长时间投入的游戏之间的重要区别。
A database showing 100% CPU usage may still spend much of its time stalled on memory rather than executing instructions.数据库显示 CPU 使用率为100%时,处理器可能仍有大量时间在等待内存,而不是执行指令。
A database showing 100% CPU usage may still spend much of its time stalled on memory rather than executing instructions. Modern CPUs can process roughly 4–8 instructions per nanosecond, while a RAM access may take 50–100 nanoseconds. That gap means a processor can potentially execute 400–800 instructions during one memory access. The piece explains how caches, memory stalls, and radix partitioning affect operations such as GROUP BY.数据库显示 CPU 使用率为100%时,处理器可能仍有大量时间在等待内存,而不是执行指令。现代 CPU 每纳秒大约可处理4至8条指令,而一次 RAM 访问可能需要50至100纳秒。这意味着处理器在等待一次内存访问期间,理论上可以执行400至800条指令。文章解释了缓存、内存停顿和基数分区如何影响 GROUP BY 等操作。
Google says it dismantled a backdoor operated by one of its manufacturing partners.谷歌称,其已拆除一家制造合作伙伴植入的后门。
Google says it dismantled a backdoor operated by one of its manufacturing partners. The backdoor was used to create a botnet containing 1.3 million Android TV devices, and Google says the manufacturer intentionally hijacked the devices with malware.谷歌称,其已拆除一家制造合作伙伴植入的后门。该后门被用于组建包含 130 万台 Android TV 设备的僵尸网络;谷歌表示,这家制造商是有意通过恶意软件劫持这些设备。
Toit v2.0.0 has been released after more than four years of alpha development.经过四年多的 alpha 开发,Toit v2.0.0 正式发布。
Toit v2.0.0 has been released after more than four years of alpha development. The project accumulated more than 2,000 commits, over 500,000 changed lines, and more than 200 alpha releases. Toit began as closed-source software, later became open source, and continued to receive volunteer contributions after Toitware shut down in 2024.经过四年多的 alpha 开发,Toit v2.0.0 正式发布。项目累计完成超过 2,000 次提交、修改超过 50 万行代码,并经历了 200 多个 alpha 版本。Toit 最初是闭源软件,后来转为开源;Toitware 于 2024 年停止运营后,项目仍由志愿者持续维护。
Researchers describe an attack in which malware on an air-gapped computer generates radio signals through memory buses.研究人员描述了一种攻击:气隙计算机上的恶意软件通过内存总线生成无线电信号。
Researchers describe an attack in which malware on an air-gapped computer generates radio signals through memory buses. The signals can encode files, images, keystrokes, biometric data, and encryption keys, which an attacker can intercept using software-defined radio and an inexpensive antenna. The paper presents the attack's design, implementation, and evaluation.研究人员描述了一种攻击:气隙计算机上的恶意软件通过内存总线生成无线电信号。这些信号可以编码文件、图像、键盘记录、生物特征数据和加密密钥,攻击者则可使用软件定义无线电和廉价天线进行拦截。论文介绍了该攻击的设计、实现和评估结果。
The paper turns the Platonic Representation Hypothesis into a theorem for separately trained JEPAs across multiple modalities, showing that they learn the same representation up to a rotation under specified positive-pair and misalignment conditions.论文将柏拉图表征假说转化为一个定理:在满足特定正样本配对和错位条件时,分别训练的多模态 JEPA 会学习到除旋转外相同的表征。
The paper turns the Platonic Representation Hypothesis into a theorem for separately trained JEPAs across multiple modalities, showing that they learn the same representation up to a rotation under specified positive-pair and misalignment conditions. It characterizes next-token prediction as a JEPA with “bad views,” explaining why language–vision alignment depends on scale. Experiments spanning 145,000 captions and the text of 940 million web images report that good views need 42 times fewer samples than next-token prediction to reach the same alignment.论文将柏拉图表征假说转化为一个定理:在满足特定正样本配对和错位条件时,分别训练的多模态 JEPA 会学习到除旋转外相同的表征。论文将下一个词预测描述为使用“坏视图”的 JEPA,并据此解释语言与视觉为何需要依靠规模才能对齐。覆盖 14.5 万条图像说明文字和 9.4 亿张网页图片文本的实验显示,达到相同对齐程度时,良好视图所需的训练样本少 42 倍。
GPT-Rosalind-6 is described as a frontier reasoning model for biology, drug discovery, and translational medicine.GPT-Rosalind-6 被描述为面向生物学、药物发现和转化医学的前沿推理模型。
GPT-Rosalind-6 is described as a frontier reasoning model for biology, drug discovery, and translational medicine. It is available only to qualified users through trusted access, with applications, security controls, and governance practices reviewed before approval. The model shares weights with GPT-6 Astra, supports some dual-use biological research, and continues to refuse biological-weaponization requests.GPT-Rosalind-6 被描述为面向生物学、药物发现和转化医学的前沿推理模型。它仅通过受信访问向合格用户提供,申请用途、安全控制和治理措施会在批准前接受审查。该模型与 GPT-6 Astra 使用相同权重,支持部分双重用途生物研究,但仍拒绝生物武器化请求。
This project collects V100 SXM2 pin maps, two- and four-GPU NVLink connection plans, and connector-only KiCad resources.该项目整理了 V100 SXM2 引脚图、双 GPU 和四 GPU 的 NVLink 连接方案,以及仅含连接器的 KiCad 资源。
This project collects V100 SXM2 pin maps, two- and four-GPU NVLink connection plans, and connector-only KiCad resources. It aims to support an open-source carrier-board design for Nvidia V100 SXM2 modules, while enabling others to develop their own designs.该项目整理了 V100 SXM2 引脚图、双 GPU 和四 GPU 的 NVLink 连接方案,以及仅含连接器的 KiCad 资源。项目目标是支持 Nvidia V100 SXM2 模块的开源载板设计,同时让其他人能够据此开发自己的设计。
Gartner estimates that nine out of ten agent-workflow steps are routine and can be handled at parity by open-weight models.Gartner 估计,智能体工作流中十分之九的步骤属于常规任务,开放权重模型可以达到同等效果。
Gartner estimates that nine out of ten agent-workflow steps are routine and can be handled at parity by open-weight models. The firm says these models cost about one-quarter as much as the cheapest closed model, while model makers may compete for the remaining 10% of more valuable steps. The analysis argues that agent harnesses may be more valuable than the underlying models.Gartner 估计,智能体工作流中十分之九的步骤属于常规任务,开放权重模型可以达到同等效果。该机构称,这些模型的成本约为最便宜闭源模型的四分之一,而模型厂商可能会争夺剩余 10% 的高价值步骤。分析认为,智能体运行框架可能比底层模型更有价值。
A team led by Dirk Witthaut at Forschungszentrum Jülich found that the brief demand spike seen on power grids every minute is caused by regular power checks from large battery-storage systems.Forschungszentrum Jülich 的 Dirk Witthaut 团队发现,电网每分钟出现的短暂需求尖峰源于大型电池储能系统定期进行的功率检查。
A team led by Dirk Witthaut at Forschungszentrum Jülich found that the brief demand spike seen on power grids every minute is caused by regular power checks from large battery-storage systems. Tennet Germany had observed the unusually regular frequency dips across the European grid for years, without initially knowing their cause. The finding could help make electricity systems more robust.Forschungszentrum Jülich 的 Dirk Witthaut 团队发现,电网每分钟出现的短暂需求尖峰源于大型电池储能系统定期进行的功率检查。德国输电运营商 Tennet 多年来一直观察到欧洲电网频率出现这种异常规律的下降,但最初并不清楚原因。这一发现可能有助于提升供电系统的稳健性。
The guide describes reducing NixOS's RAM requirements to 512 MB.这份指南介绍了如何将 NixOS 的内存需求降至 512 MB。
The guide describes reducing NixOS's RAM requirements to 512 MB. The author expected the work to take half a day but spent nearly three days on it, and notes that the guide focuses on RAM rather than disk-space reduction. Combining the approaches may introduce defaults that users do not expect.这份指南介绍了如何将 NixOS 的内存需求降至 512 MB。作者原本预计半天完成,但实际花了近三天,并说明文章重点是降低内存占用,而不是节省磁盘空间。将两类优化结合使用,可能会引入终端用户意料之外的默认设置。
Researchers revisited 50 published findings about language models across 20 models.研究人员在 20 个模型上重新检验了 50 项已发表的语言模型研究发现。
Researchers revisited 50 published findings about language models across 20 models. Within 15 months of publication, most tested claims either no longer held or had substantially weaker effects. The results suggest that some empirical knowledge about AI may have a short shelf life.研究人员在 20 个模型上重新检验了 50 项已发表的语言模型研究发现。发表后 15 个月内,大多数被测试的结论要么不再成立,要么效果明显减弱。结果表明,部分关于 AI 的经验性认识可能具有较短的有效期。
Rustable is an open-source, native database client for macOS, Windows, and Linux.Rustable 是一款面向 macOS、Windows 和 Linux 的开源原生数据库客户端,使用 Rust 编写。
Rustable is an open-source, native database client for macOS, Windows, and Linux. It is written in Rust.Rustable 是一款面向 macOS、Windows 和 Linux 的开源原生数据库客户端,使用 Rust 编写。
The project runs the original Windows release of ReBirth RB-338 in a browser using Wine and WebAssembly.这个项目利用 Wine 和 WebAssembly,在浏览器中运行原版 Windows 版 ReBirth RB-338。
The project runs the original Windows release of ReBirth RB-338 in a browser using Wine and WebAssembly.这个项目利用 Wine 和 WebAssembly,在浏览器中运行原版 Windows 版 ReBirth RB-338。
Experts warn that AI's dominance of the U.S.专家警告,AI 对美国股市的主导地位正在制造新的风险。
Experts warn that AI's dominance of the U.S. stock market is creating its own risks. The discussion points to the concentration of some of the world's wealthiest people around President Donald Trump at the White House AI Summit.专家警告,AI 对美国股市的主导地位正在制造新的风险。相关讨论提到,在白宫 AI 峰会上,全球一些最富有的人聚集在美国总统唐纳德·特朗普周围。
Two brothers launched a small UK SaaS business in 2006 using dial-up modems and basic HTML knowledge.两兄弟于 2006 年使用拨号调制解调器和基础 HTML 知识创办了一家英国小型 SaaS 公司。
Two brothers launched a small UK SaaS business in 2006 using dial-up modems and basic HTML knowledge. After 20 years, the service has completed 766,615,348 file conversions. The account reflects on the ups and downs of running the business without outside funding.两兄弟于 2006 年使用拨号调制解调器和基础 HTML 知识创办了一家英国小型 SaaS 公司。20 年后,这项服务已经完成 766,615,348 次文件转换。文章回顾了不依赖外部融资经营这家公司的起伏。
It points out that symbols such as ™, ℠, ℞, ℡, and № are individual Unicode characters, and notes that .tel and .no are also top-level domains.文章介绍了使用罕见 Unicode 字符制作域名变体的做法。
It points out that symbols such as ™, ℠, ℞, ℡, and № are individual Unicode characters, and notes that .tel and .no are also top-level domains.文章介绍了使用罕见 Unicode 字符制作域名变体的做法。它指出 ™、℠、℞、℡ 和 № 等符号都各自是单个 Unicode 字符,并注意到 .tel 和 .no 也都是顶级域名。
The essay recounts how paradoxes such as Russell's paradox triggered a foundational crisis in mathematics.文章回顾了罗素悖论等悖论如何引发数学基础危机。
The essay recounts how paradoxes such as Russell's paradox triggered a foundational crisis in mathematics. Hilbert sought a complete and consistent axiomatic foundation, but Gödel's incompleteness theorems showed that no complete, consistent, self-supporting system of mathematics can exist. The title quotes Hilbert's declaration, “We must know—we will know.”文章回顾了罗素悖论等悖论如何引发数学基础危机。希尔伯特曾试图建立完备且一致的公理体系,但哥德尔不完备定理表明,不可能存在一个完备、一致且自我支撑的数学体系。标题引用了希尔伯特的宣言:“我们必须知道——我们必将知道。”
One example is its turquoise roof: the roof is made of copper, whose reaction with oxygen, water, and carbon dioxide gradually forms a green patina.文章通过一百多个细节,讲述汉堡市政厅如何呈现这座城市的历史。
One example is its turquoise roof: the roof is made of copper, whose reaction with oxygen, water, and carbon dioxide gradually forms a green patina.文章通过一百多个细节,讲述汉堡市政厅如何呈现这座城市的历史。其中一个例子是它的蓝绿色屋顶:屋顶由铜制成,铜与空气中的氧气、水和二氧化碳发生反应,经过多年逐渐形成绿色的铜锈。
The essay explains projective geometry as a natural response to practical problems rather than an arbitrary abstraction.这篇文章将射影几何解释为解决实际问题的自然工具,而非任意的抽象。
The essay explains projective geometry as a natural response to practical problems rather than an arbitrary abstraction. Projective space and homogeneous coordinates eliminate exceptional cases and informal references to points at infinity. The author describes coming to appreciate these ideas only later, after initially finding them unmotivated as a student.这篇文章将射影几何解释为解决实际问题的自然工具,而非任意的抽象。射影空间和齐次坐标能够消除特殊情况,以及对“无穷远处对象”的非严格描述。作者回忆说,自己学生时代觉得这些概念缺乏动机,后来才认识到它们的自然性。
NASA's Neil Gehrels Swift Observatory, which has been in orbit for 22 years, is expected to reenter the atmosphere tomorrow.NASA 的 Neil Gehrels Swift 天文台已在轨运行 22 年,预计将于明天重返大气层。
NASA's Neil Gehrels Swift Observatory, which has been in orbit for 22 years, is expected to reenter the atmosphere tomorrow. Plans to save the defunct space telescope failed, and its descent can be tracked.NASA 的 Neil Gehrels Swift 天文台已在轨运行 22 年,预计将于明天重返大气层。拯救这台失效太空望远镜的计划未能成功,目前可以追踪其下降过程。
A 2024 National Academies report identified long-term zinc creep in cable spelter sockets as the root cause of the Arecibo telescope’s 2020 collapse.美国国家科学院 2024 年的一份报告认定,电缆锌制套筒长期发生锌蠕变,是阿雷西博望远镜 2020 年倒塌的根本原因。
A 2024 National Academies report identified long-term zinc creep in cable spelter sockets as the root cause of the Arecibo telescope’s 2020 collapse. The failure had not previously been documented in this form, and the report draws lessons for other unique critical science facilities. The sequence began with Hurricane Maria more than three years before the collapse.美国国家科学院 2024 年的一份报告认定,电缆锌制套筒长期发生锌蠕变,是阿雷西博望远镜 2020 年倒塌的根本原因。这种失效方式此前从未被记录,报告还为其他独特且关键的科学设施总结了经验。事故过程始于倒塌三年多前的玛丽亚飓风。
The essay revisits the Exponential Time Hypothesis and Strong Exponential Time Hypothesis.这篇文章重新审视指数时间假说和强指数时间假说。
The essay revisits the Exponential Time Hypothesis and Strong Exponential Time Hypothesis. It describes them as deliberately provocative working hypotheses rather than conjectures, intended both to challenge claims that P equals NP and to explore relationships among problems when SAT requires far more than polynomial time. Earlier work had improved brute-force exponents, but the connections between those results were unclear.这篇文章重新审视指数时间假说和强指数时间假说。文章称,它们原本是带有挑衅性的工作假说,而非严格猜想:一方面用于挑战“P=NP”的主张,另一方面用于探索当 SAT 的真实复杂度远超多项式时间时,不同问题之间会如何关联。此前人们不断改进暴力算法的指数常数,但这些结果之间的联系并不清晰。
Vathography is an art practice that turns the depth estimated from a photograph into an artwork.Vathography 是一种将照片估算出的深度转化为艺术作品的创作实践。
Vathography is an art practice that turns the depth estimated from a photograph into an artwork. A diffusion transformer infers the relative distance of every point in a scene from a single image, translating the photograph's visual data into volume rather than documentary imagery. The work was inspired by Marigold V2's ability to produce detailed component maps suitable for high-resolution art.Vathography 是一种将照片估算出的深度转化为艺术作品的创作实践。扩散 Transformer 根据单张图像推断场景中每个点的相对距离,把照片数据转译为体积和形态,而不再保留其作为纪实照片的性质。Marigold V2 能生成足够细致、适合高分辨率创作的分层图像,促成了这一探索。
PyWolf3D is a Wolfenstein 3D engine and game clone written from scratch in Python using Pygame.PyWolf3D 是一个使用 Python 和 Pygame 从零编写的 Wolfenstein 3D 引擎及游戏克隆版,并提供演示。
PyWolf3D is a Wolfenstein 3D engine and game clone written from scratch in Python using Pygame. A demo is included.PyWolf3D 是一个使用 Python 和 Pygame 从零编写的 Wolfenstein 3D 引擎及游戏克隆版,并提供演示。
GUI64 is a graphical layer running on top of the Commodore 64's built-in Kernal rather than a standalone operating system.GUI64 是运行在 Commodore 64 内置 Kernal 之上的图形层,而不是独立操作系统。
GUI64 is a graphical layer running on top of the Commodore 64's built-in Kernal rather than a standalone operating system. It provides movable windows for managing floppy-disk files and SD2IEC disk images, and can mount images directly on Ultimate devices. The current version is reported as 2.1 and supports a mouse, joystick, and mouse wheel through a Micromys adapter or the VICE emulator.GUI64 是运行在 Commodore 64 内置 Kernal 之上的图形层,而不是独立操作系统。它通过可移动窗口管理软盘文件和 SD2IEC 磁盘镜像,并可在 Ultimate 设备中直接挂载镜像。据资料显示,当前版本为 2.1,支持鼠标、摇杆,以及通过 Micromys 适配器或 VICE 模拟器使用鼠标滚轮。
Phoria2 is a small JavaScript library for web-based 3D charts, labeled scenes, dashboards, and product viewers, using a WebGL 2 renderer.Phoria2 是一个基于 WebGL 2 渲染器的轻量 JavaScript 3D 库,可用于网页图表、带标签场景、仪表盘和产品查看器。
Phoria2 is a small JavaScript library for web-based 3D charts, labeled scenes, dashboards, and product viewers, using a WebGL 2 renderer. Its API is designed to be learned quickly and is intended for both humans and coding agents. The project rewrites phoria.js, a 2013 library that rendered 3D through the 2D canvas context.Phoria2 是一个基于 WebGL 2 渲染器的轻量 JavaScript 3D 库,可用于网页图表、带标签场景、仪表盘和产品查看器。它的 API 设计为易于快速学习,并同时面向人类开发者和编码代理。该项目重写了 2013 年通过 2D canvas 上下文渲染 3D 的 phoria.js。
PsiQuantum unveiled optical circuit switching technology based on its low-loss silicon-photonics platform.PsiQuantum 发布了基于低损耗硅光子平台的光学电路交换技术。
PsiQuantum unveiled optical circuit switching technology based on its low-loss silicon-photonics platform. The company presents it as a response to the growing networking demands created by large AI compute clusters. Optical connectivity is increasingly being considered for faster, more efficient, and more scalable data movement between computing systems.PsiQuantum 发布了基于低损耗硅光子平台的光学电路交换技术。该公司将其定位为应对大型 AI 计算集群不断增长的网络需求。随着计算系统之间的数据传输成为 AI 扩展的瓶颈,光互连正被越来越多地用于实现更快、更高效且更易扩展的连接。
Radiation Map of the World is a community-measured global radiation map.《世界辐射地图》是一张由社区共同测量的全球辐射地图。
Radiation Map of the World is a community-measured global radiation map. It tracks detector-recorded routes and areas and provides live statistics from RadiaCode devices around the world.《世界辐射地图》是一张由社区共同测量的全球辐射地图。它记录探测器采集的路线和区域,并提供来自世界各地 RadiaCode 设备的实时统计数据。
Researchers developed and tested a three-dimensional thermal cloak that can conceal complex objects from infrared cameras while protecting them from extreme temperatures.研究人员开发并测试了一种三维热隐身斗篷,能够让复杂物体避开红外摄像机的探测,同时保护其免受极端温度影响。
Researchers developed and tested a three-dimensional thermal cloak that can conceal complex objects from infrared cameras while protecting them from extreme temperatures. Rather than blocking heat, it guides heat around the hidden object and works across different heat directions and object shapes. Potential applications include electronics cooling, thermal protection, and reducing thermal detection.研究人员开发并测试了一种三维热隐身斗篷,能够让复杂物体避开红外摄像机的探测,同时保护其免受极端温度影响。它不是阻挡热量,而是将热量引导至物体周围,并能适应不同的热流方向和物体形状。潜在应用包括电子设备散热、热防护和降低热特征探测。
Mandiant describes LLMJacking attacks in which stolen AI credentials generate fraudulent cloud bills.Mandiant 描述了利用被盗 AI 凭证制造虚假云账单的 LLMJacking 攻击。
Mandiant describes LLMJacking attacks in which stolen AI credentials generate fraudulent cloud bills. The attacks can cost victims hundreds of thousands of dollars, while underground markets trade compromised credentials for access to hosted language models.Mandiant 描述了利用被盗 AI 凭证制造虚假云账单的 LLMJacking 攻击。这类攻击可能让受害者损失数十万美元,地下市场则交易被入侵的凭证,以访问托管的大语言模型。
The interview covers building a search engine in Rust, competing with Lucene, maintaining open source, and lessons from Tantivy's development.这次访谈讨论了用 Rust 构建搜索引擎、与 Lucene 竞争、维护开源项目,以及 Tantivy 开发过程中的经验。
The interview covers building a search engine in Rust, competing with Lucene, maintaining open source, and lessons from Tantivy's development. Tantivy began in 2017 as a small modular Rust project and later became the engine behind Quickwit, ParadeDB, LNX, and other search products. Its development also prompted performance collaboration with the Lucene team.这次访谈讨论了用 Rust 构建搜索引擎、与 Lucene 竞争、维护开源项目,以及 Tantivy 开发过程中的经验。Tantivy 于 2017 年作为一个小型模块化 Rust 项目出现,后来成为 Quickwit、ParadeDB、LNX 等搜索产品的引擎。其发展还促成了与 Lucene 团队的性能合作。
This collection catalogs all 1,504 marks from Yasaburo Kuwayama's 1973 book, with designer, company, and chapter information.这个资料库收录了 Yasaburo Kuwayama 1973 年著作中的全部 1,504 个商标,并提供设计师、公司和章节信息。
This collection catalogs all 1,504 marks from Yasaburo Kuwayama's 1973 book, with designer, company, and chapter information. The marks range from human figures and birds to arrows and hiragana. The collection is available as free SVG and PNG files for noncommercial use.这个资料库收录了 Yasaburo Kuwayama 1973 年著作中的全部 1,504 个商标,并提供设计师、公司和章节信息。图案涵盖人物、鸟类、箭头和日文平假名等类型。所有内容均以 SVG 和 PNG 格式免费提供,限非商业使用。
The study asked five LLMs to modernize 100 WordPress 4.0 PHP files to PHP 8.3 without project context or execution feedback.该研究让五个 LLM 在没有项目上下文和运行反馈的情况下,将 100 个 WordPress 4.0 PHP 文件现代化到 PHP 8.3。
The study asked five LLMs to modernize 100 WordPress 4.0 PHP files to PHP 8.3 without project context or execution feedback. The untouched baseline had 83 files passing all hidden checks, while the best LLM left 59; models broke 32–51 working files and repaired at most 10. Rector performed better on compatibility, and many regressions came from type declarations that PHP rejected.该研究让五个 LLM 在没有项目上下文和运行反馈的情况下,将 100 个 WordPress 4.0 PHP 文件现代化到 PHP 8.3。未修改的基线有 83 个文件通过全部隐藏检查,而表现最好的 LLM 只剩 59 个;各模型破坏了 32 至 51 个原本可运行的文件,最多只修复 10 个。Rector 在兼容性方面表现更好,许多回归源于 PHP 本身拒绝的类型声明。
WiCi is a systems research lab focused on wireless GPU computing and local AI infrastructure.WiCi 是一个专注于无线 GPU 计算和本地 AI 基础设施的系统研究实验室。
WiCi is a systems research lab focused on wireless GPU computing and local AI infrastructure. It develops systems for sharing compute across devices while keeping models and data under local control. Its work targets local inference and wireless infrastructure for advanced AI.WiCi 是一个专注于无线 GPU 计算和本地 AI 基础设施的系统研究实验室。它开发让多台设备共享计算资源、同时保持模型和数据本地可控的系统。研究方向包括先进 AI 的本地推理和无线基础设施。