{
  "title": "2026-09-22 AI日更 | GPT-6 Astra 一天上线视频功能，智能体同日进入芯片设计与精神科质控",
  "url": "https://miaok.ong/ai-daily/ai-daily-2026-09-22/",
  "date": "2026-09-22T07:00:00+08:00",
  "lastmod": "2026-09-22T07:00:00+08:00",
  "type": "ai-daily",
  "kind": "page",
  "language": "zh",
  "description": "今日 AI 主线从能力发布转向交付与治理并进：OpenAI 同日成立数学顾问组、发布全球标准路线图并扩展 Academy，GPT-6 Astra 已支持 Higgsfield 一天上线视频功能；智能体进入芯片设计与精神科质控等专业场景。Grok 代理团队提交 2500 个 PR，引发工程可靠性与审查负担争议。",
  "keywords": null,
  "tags": [],
  "categories": [],
  "author": "孔淼",
  "image": "https://miaok.ong/images/avatar.jpg",
  "content": "\u003ch1 id=\"2026-09-22-ai日更--gpt-6-astra-一天上线视频功能智能体同日进入芯片设计与精神科质控\"\u003e\n  2026-09-22 AI日更 | GPT-6 Astra 一天上线视频功能，智能体同日进入芯片设计与精神科质控\n  \u003ca class=\"heading-link\" href=\"#2026-09-22-ai%e6%97%a5%e6%9b%b4--gpt-6-astra-%e4%b8%80%e5%a4%a9%e4%b8%8a%e7%ba%bf%e8%a7%86%e9%a2%91%e5%8a%9f%e8%83%bd%e6%99%ba%e8%83%bd%e4%bd%93%e5%90%8c%e6%97%a5%e8%bf%9b%e5%85%a5%e8%8a%af%e7%89%87%e8%ae%be%e8%ae%a1%e4%b8%8e%e7%b2%be%e7%a5%9e%e7%a7%91%e8%b4%a8%e6%8e%a7\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h1\u003e\n\u003cblockquote\u003e\n\u003cp\u003e今日 AI 主线从能力发布转向交付与治理并进：OpenAI 同日成立数学顾问组、发布全球标准路线图并扩展 Academy，GPT-6 Astra 已支持 Higgsfield 一天上线视频功能；智能体进入芯片设计与精神科质控等专业场景。Grok 代理团队提交 2500 个 PR，引发工程可靠性与审查负担争议。\u003c/p\u003e\n\u003c/blockquote\u003e\n\u003ch2 id=\"-本期-watch-list-深度导读\"\u003e\n  📖 本期 Watch List 深度导读\n  \u003ca class=\"heading-link\" href=\"#-%e6%9c%ac%e6%9c%9f-watch-list-%e6%b7%b1%e5%ba%a6%e5%af%bc%e8%af%bb\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cp\u003e今日有三条阅读主线。一是 OpenAI 治理与落地密集更新：数学与 AI 顾问组、全球标准路线图、Academy 新学习路径，以及 Higgsfield 借助 GPT-6 Astra 一天上线视频功能，安全、教育与商用生态同步推进。二是 arXiv 上长上下文与注意力效率研究集中出现：半径有界稀疏 prefill、MoE 路由与注意力耦合、从注意力拓扑检测幻觉等，工程团队值得深读。三是智能体进入专业深水区：芯片设计探索更高层抽象、精神科接诊引入临床质控，Naveen Rao 播客则从能源墙角度讨论算力成本。\u003c/p\u003e\n\u003ch2 id=\"-x-平台-ai-热点快讯\"\u003e\n  🌐 X 平台 AI 热点快讯\n  \u003ca class=\"heading-link\" href=\"#-x-%e5%b9%b3%e5%8f%b0-ai-%e7%83%ad%e7%82%b9%e5%bf%ab%e8%ae%af\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003ch3 id=\"话题-1zai-open-sources-zcode-after-secret-code-uploads\"\u003e\n  话题 1:Z.ai Open-Sources ZCode After Secret Code Uploads\n  \u003ca class=\"heading-link\" href=\"#%e8%af%9d%e9%a2%98-1zai-open-sources-zcode-after-secret-code-uploads\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e分类:AI · News\u003c/li\u003e\n\u003cli\u003e概况:热度时间:21 hours ago,相关帖子数:3200\u003c/li\u003e\n\u003cli\u003e是什么事:Z.ai 在 ZCode 代码被秘密上传后，宣布将 ZCode 开源。\u003c/li\u003e\n\u003cli\u003e为什么重要:这一事件凸显了 AI 模型代码的泄露与开源边界问题，可能影响代码生成模型的透明度、安全性和后续社区采用。\u003c/li\u003e\n\u003cli\u003e讨论概况:X 上的讨论集中在 Z.ai 开源是否为被动应对、秘密上传对模型安全的影响，以及 ZCode 的实际能力与许可证限制。\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"话题-2spacexai-engineer-ships-2500-prs-using-grok-bot-agent-teams\"\u003e\n  话题 2:SpaceXAI Engineer Ships 2,500 PRs Using Grok Bot Agent Teams\n  \u003ca class=\"heading-link\" href=\"#%e8%af%9d%e9%a2%98-2spacexai-engineer-ships-2500-prs-using-grok-bot-agent-teams\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e分类:AI · News\u003c/li\u003e\n\u003cli\u003e概况:热度时间:7 hours ago,相关帖子数:1200\u003c/li\u003e\n\u003cli\u003e是什么事:一名 SpaceX AI 工程师使用 Grok 机器人代理团队提交了 2,500 个代码拉取请求（PR）。\u003c/li\u003e\n\u003cli\u003e为什么重要:该事件展示了 AI 智能体在规模化软件工程中的实际应用能力，可能推动软件开发流程向人机协同自动化转变。\u003c/li\u003e\n\u003cli\u003e讨论概况:X 上的讨论聚焦于这些 PR 的真实质量与可维护性，争议点包括自动化提交是否加重代码审查负担、效率数据是否可信，以及该模式能否在关键生产环境中推广。\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"话题-3ling-30-flash-vl-ai-transforms-product-photos-into-marketing-plans-and-prototypes\"\u003e\n  话题 3:Ling-3.0-flash-VL AI Transforms Product Photos into Marketing Plans and Prototypes\n  \u003ca class=\"heading-link\" href=\"#%e8%af%9d%e9%a2%98-3ling-30-flash-vl-ai-transforms-product-photos-into-marketing-plans-and-prototypes\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e分类:AI · News\u003c/li\u003e\n\u003cli\u003e概况:热度时间:13 hours ago,相关帖子数:1200\u003c/li\u003e\n\u003cli\u003e是什么事:Ling-3.0-flash-VL 发布，能够将产品照片自动转化为营销计划和产品原型。\u003c/li\u003e\n\u003cli\u003e为什么重要:该技术展示了多模态视觉语言模型在商业创意生成与设计流程自动化上的新应用，可能降低营销与原型制作门槛。\u003c/li\u003e\n\u003cli\u003e讨论概况:X 上的讨论集中在生成质量与实际可用性，以及它是否会取代营销和设计人员，还是作为辅助工具。\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch4 id=\"今日-x-上的-ai-舆情小结\"\u003e\n  今日 X 上的 AI 舆情小结\n  \u003ca class=\"heading-link\" href=\"#%e4%bb%8a%e6%97%a5-x-%e4%b8%8a%e7%9a%84-ai-%e8%88%86%e6%83%85%e5%b0%8f%e7%bb%93\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h4\u003e\n\u003cp\u003e今日舆论主线围绕 AI 智能体和多模态模型加速进入实际生产环节展开，从代码生成、规模化软件工程到营销设计自动化均有体现。共识在于这些工具能降低门槛、提升效率，并推动人机协同自动化；但分歧也很明显：Z.ai 开源被视为泄露后的被动应对，Grok 提交的大量 PR 被质疑质量与审查负担，Ling-3.0-flash-VL 则引发“取代还是辅助”的争论。潜在风险集中在模型代码泄露与开源边界不清、自动化产出在关键生产环境中的可靠性不足，以及营销、设计等岗位可能受到的冲击。\u003c/p\u003e\n\u003ch2 id=\"-大佬观点influencer-insights\"\u003e\n  💡 大佬观点(Influencer Insights)\n  \u003ca class=\"heading-link\" href=\"#-%e5%a4%a7%e4%bd%ac%e8%a7%82%e7%82%b9influencer-insights\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cblockquote\u003e\n\u003cp\u003e今日大佬观点暂缺,推荐阅读 Watch List 深度内容。\u003c/p\u003e\n\u003c/blockquote\u003e\n\u003ch2 id=\"-附录今日-watch-list-更新源列表\"\u003e\n  📚 附录:今日 Watch List 更新源列表\n  \u003ca class=\"heading-link\" href=\"#-%e9%99%84%e5%bd%95%e4%bb%8a%e6%97%a5-watch-list-%e6%9b%b4%e6%96%b0%e6%ba%90%e5%88%97%e8%a1%a8\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h2\u003e\n\u003cblockquote\u003e\n\u003cp\u003e时间窗口:最近 3 天;覆盖 22 个源;共 37 条更新\u003c/p\u003e\n\u003c/blockquote\u003e\n\u003ch3 id=\"all-in-podcast-a_full\"\u003e\n  All-In Podcast (A_full)\n  \u003ca class=\"heading-link\" href=\"#all-in-podcast-a_full\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003e\u003ca href=\"https://allinchamathjason.libsyn.com/naveen-rao-4d-computing-ais-energy-wall-beating-biology\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eNaveen Rao: 4D Computing, AI\u0026rsquo;s Energy Wall \u0026amp; Beating Biology\u003c/a\u003e\u003c/strong\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-22 05:03 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- The cost of a token, power contracts \u0026amp; the gap to close.\n\u003cul\u003e\n\u003cli\u003eIREN is a vertically integrated AI Cloud platform, delivering data centers, compute and software for AI training and inference.\u003c/li\u003e\n\u003cli\u003eOracle connects the data, applications, and infrastructure that turn AI into business outcomes—with the flexibility, choice, and control to optimize as AI evolves.\u003c/li\u003e\n\u003cli\u003eEY helps tech innovators scale from startup to exit to megacap.\u003c/li\u003e\n\u003cli\u003eWe\u0026rsquo;re focused on giving every person the tools to reach their full potential and making sure the benefits of technology are distributed to all.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003e(0:00) Welcome Naveen Rao\u003c/li\u003e\n\u003cli\u003e(4:45) Is energy really the problem\u003c/li\u003e\n\u003cli\u003eThe cost of a token, power contracts \u0026amp; the gap to close\u003c/li\u003e\n\u003cli\u003e(10:24) Cutting out the middleman: abstractions, dynamical systems \u0026amp; a new kind of machine\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"stratechery-by-ben-thompson-a_full\"\u003e\n  Stratechery by Ben Thompson (A_full)\n  \u003ca class=\"heading-link\" href=\"#stratechery-by-ben-thompson-a_full\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\u003cstrong\u003e\u003ca href=\"https://stratechery.com/2026/frontier-overhangs/\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eFrontier Overhangs\u003c/a\u003e\u003c/strong\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 18:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- \u003cstrong\u003eListen to this \u003cstrong\u003epost\u003c/strong\u003e:\u003c/strong\u003e.\n\u003cul\u003e\n\u003cli\u003eThere has been, over the last week, what I think is a healthy debate about the philosophy and psychology that undergirds the views of meaningful segments of the AI community, particularly those obsessed with doomsday scenarios.\u003c/li\u003e\n\u003cli\u003eIt is, in the end, difficult to reason with a philosophy that grants equivalent moral weight to not just all beings — human or not — who exist today, but who may ever exist in the future; this tilts the scales in such an absurd fashion towards safetyism that innovation is impossible and freedom is intolerable.\u003c/li\u003e\n\u003cli\u003eWorse, it taps into the psychology of religion, where dissent is not brooked and questioning the premise is heresy.\u003c/li\u003e\n\u003cli\u003eThis analysis is in a similar vein but from the opposite direction: “Pacing the Frontier” is framed as — and I believe motivated by — a concern about safety, but it also happens to address several distinct problems faced by the frontier labs.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003eListen to this post :\u003c/li\u003e\n\u003cli\u003eLog in to listen\u003c/li\u003e\n\u003cli\u003eThere has been, over the last week, what I think is a healthy debate about the philosophy and psychology that undergirds the views of meaningful segments of the…\u003c/li\u003e\n\u003cli\u003eIt is, in the end, difficult to reason with a philosophy that grants equivalent moral weight to not just all beings — human or not — who exist today, but who ma…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"openai-blog-a_full\"\u003e\n  OpenAI Blog (A_full)\n  \u003ca class=\"heading-link\" href=\"#openai-blog-a_full\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://openai.com/index/advisory-group-on-mathematics-and-ai\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eAdvisory Group on Mathematics and Artificial Intelligence\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 20:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- OpenAI is working with an independent Advisory Group on Mathematics and Artificial Intelligence to guide the review and communication of emerging AI results.\n\u003cul\u003e\n\u003cli\u003eThis piece from OpenAI Blog explains how Advisory Group on Mathematics and Artificial Intelligence shapes the broader AI and infrastructure landscape.\u003c/li\u003e\n\u003cli\u003eIt also surfaces practical implications for founders, operators, and investors following Advisory Group on Mathematics and Artificial Intelligence.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003eOpenAI is working with an independent Advisory Group on Mathematics and Artificial Intelligence to guide the review and communication of emerging AI results.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://openai.com/index/higgsfield-from-prompt-to-production-with-astra\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eHiggsfield AI ships new video features in a day with GPT-6 Astra\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 20:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- With GPT-6 Astra, Higgsfield AI makes video ad creation easier for small businesses and brings new creative tools to market faster.\n\u003cul\u003e\n\u003cli\u003eThis piece from OpenAI Blog explains how Higgsfield AI ships new video features in a day with GPT-6 Astra shapes the broader AI and infrastructure landscape.\u003c/li\u003e\n\u003cli\u003eIt also surfaces practical implications for founders, operators, and investors following Higgsfield AI ships new video features in a day with GPT-6 Astra.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003eWith GPT-6 Astra, Higgsfield AI makes video ad creation easier for small businesses and brings new creative tools to market faster.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://openai.com/index/building-standards-next-phase-ai\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eBuilding standards for the next phase of AI\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 18:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- OpenAI outlines a path to shared global AI standards, calling for coordinated evaluation, reporting, and governance to improve safety.\n\u003cul\u003e\n\u003cli\u003eThis piece from OpenAI Blog explains how Building standards for the next phase of AI shapes the broader AI and infrastructure landscape.\u003c/li\u003e\n\u003cli\u003eIt also surfaces practical implications for founders, operators, and investors following Building standards for the next phase of AI.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003eOpenAI outlines a path to shared global AI standards, calling for coordinated evaluation, reporting, and governance to improve safety.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://openai.com/index/expanding-openai-academy-with-new-learning-paths\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eExpanding OpenAI Academy with new learning paths\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 15:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- Explore new OpenAI Academy learning paths for employees, developers, leaders, educators, and students to build and demonstrate practical AI skills.\n\u003cul\u003e\n\u003cli\u003eThis piece from OpenAI Blog explains how Expanding OpenAI Academy with new learning paths shapes the broader AI and infrastructure landscape.\u003c/li\u003e\n\u003cli\u003eIt also surfaces practical implications for founders, operators, and investors following Expanding OpenAI Academy with new learning paths.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003eExplore new OpenAI Academy learning paths for employees, developers, leaders, educators, and students to build and demonstrate practical AI skills.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://openai.com/index/v7\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eHow V7 gives AI agents institutional memory\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 08:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- Using GPT-5.6, V7 turns scattered company files into context agents can use to complete complex, source-linked work.\n\u003cul\u003e\n\u003cli\u003eThis piece from OpenAI Blog explains how How V7 gives AI agents institutional memory shapes the broader AI and infrastructure landscape.\u003c/li\u003e\n\u003cli\u003eIt also surfaces practical implications for founders, operators, and investors following How V7 gives AI agents institutional memory.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003eUsing GPT-5.6, V7 turns scattered company files into context agents can use to complete complex, source-linked work.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"arxiv-csai-b_introsearch\"\u003e\n  ArXiv cs.AI (B_intro+search)\n  \u003ca class=\"heading-link\" href=\"#arxiv-csai-b_introsearch\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20971\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eRBS-Attention: Radius-Bounded Sparse Prefill for Long-Context Large Language Models\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20971v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Long-context large language model inference is increasingly limited by prefill, where dense self-attention processes the entire prompt before generation begins.\u003c/li\u003e\n\u003cli\u003eSparse block selection can reduce this cost, but a block centroid may hide a highly relevant token among many irrelevant ones.\u003c/li\u003e\n\u003cli\u003eWe call this failure mode mean dilution and propose RBS-Attention, a training-free sparse-prefill method with two complementary selection branches.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20971v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Long-context large language model inference is increasingly limited by prefill, where dense self-attention processes the entire prompt before generati…\u003c/li\u003e\n\u003cli\u003eSparse block selection can reduce this cost, but a block centroid may hide a highly relevant token among many irrelevant ones\u003c/li\u003e\n\u003cli\u003eWe call this failure mode mean dilution and propose RBS-Attention, a training-free sparse-prefill method with two complementary selection branches\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20974\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eAttention-Aware Routing: Coupling Routing and Attention in MoEs\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20974v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: In Mixture-of-Experts language models, the router typically selects and weights experts based on the token\u0026rsquo;s hidden state, utilizing limited contextual information.\u003c/li\u003e\n\u003cli\u003eWe propose Attention-Aware Routing (AAR), which augments the router with temporal and spectral features extracted from a sliding window of attention weights that represent a summary of the model\u0026rsquo;s contextual state, disentangled from the hidden state.\u003c/li\u003e\n\u003cli\u003eKeeping the base transformer entirely frozen, we train only the routing parameters, isolating routing as the sole variable.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20974v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: In Mixture-of-Experts language models, the router typically selects and weights experts based on the token\u0026rsquo;s hidden state, utilizing limited contextua…\u003c/li\u003e\n\u003cli\u003eWe propose Attention-Aware Routing (AAR), which augments the router with temporal and spectral features extracted from a sliding window of attention weights tha…\u003c/li\u003e\n\u003cli\u003eKeeping the base transformer entirely frozen, we train only the routing parameters, isolating routing as the sole variable\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20981\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eCaLR: Causal Latent Revision for Robust Diffusion Reasoning\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20981v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Autoregressive (AR) models suffer from local greediness, while diffusion language models (DLMs) often lack the strict causal structure required for reasoning.\u003c/li\u003e\n\u003cli\u003eTo combine the advantages and overcome the drawbacks of the dual, we propose Causal Latent Revision (CaLR), a framework that reformulates reasoning as constrained latent optimization.\u003c/li\u003e\n\u003cli\u003eBy adopting a causal topology matrix (CTM) from an expert model and implicit differentiation, CaLR performs gradient-guided ``thought revision\u0026quot; to enforce logical consistency, enabling dynamic self-correction of intermediate steps during parallel generation.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20981v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Autoregressive (AR) models suffer from local greediness, while diffusion language models (DLMs) often lack the strict causal structure required for re…\u003c/li\u003e\n\u003cli\u003eTo combine the advantages and overcome the drawbacks of the dual, we propose Causal Latent Revision (CaLR), a framework that reformulates reasoning as constrain…\u003c/li\u003e\n\u003cli\u003eBy adopting a causal topology matrix (CTM) from an expert model and implicit differentiation, CaLR performs gradient-guided ``thought revision\u0026quot; to enforce logic…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21061\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eLoRA Enhanced Contrastive Learning with SAS Vision Transformers\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21061v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Automatic target recognition (ATR) with synthetic aperture sonar (SAS) supports advanced naval capabilities, but deep learning is constrained by scarce target imagery, background clutter, and human-in-the-loop assessment.\u003c/li\u003e\n\u003cli\u003eWe adapt DINOv3 Vision Transformer (ViT) models to underwater SAS ATR using a three-stage parameter-efficient framework.\u003c/li\u003e\n\u003cli\u003eStage 1 uses Low-Rank Adaptation (LoRA) while freezing the ViT backbone, bridging the gap between natural-image pretraining and underwater acoustic propagation.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21061v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Automatic target recognition (ATR) with synthetic aperture sonar (SAS) supports advanced naval capabilities, but deep learning is constrained by scarc…\u003c/li\u003e\n\u003cli\u003eWe adapt DINOv3 Vision Transformer (ViT) models to underwater SAS ATR using a three-stage parameter-efficient framework\u003c/li\u003e\n\u003cli\u003eStage 1 uses Low-Rank Adaptation (LoRA) while freezing the ViT backbone, bridging the gap between natural-image pretraining and underwater acoustic propagation\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21096\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eDetecting Hallucination in LLMs: Tracing the Topological Signatures of Impaired Context Sharing\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21096v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: In this work, we examine the topology of information flow patterns within attention graphs to effectively distinguish hallucinated from non-hallucinated responses.\u003c/li\u003e\n\u003cli\u003eWe analyze the Forman-Ricci curvature to identify structural patterns indicating information bottlenecks in attention graphs.\u003c/li\u003e\n\u003cli\u003eWe then introduce a method that captures both semi-local and global information-flow characteristics of attention heads associated with hallucinated responses.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21096v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: In this work, we examine the topology of information flow patterns within attention graphs to effectively distinguish hallucinated from non-hallucinat…\u003c/li\u003e\n\u003cli\u003eWe analyze the Forman-Ricci curvature to identify structural patterns indicating information bottlenecks in attention graphs\u003c/li\u003e\n\u003cli\u003eWe then introduce a method that captures both semi-local and global information-flow characteristics of attention heads associated with hallucinated responses\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21113\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eDecoupling Internal Representational Changes and Causal Importance in Fine-Tuned Large Language Models\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21113v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Fine-tuning has emerged as a widely adopted approach for adapting LLMs to a variety of downstream tasks.\u003c/li\u003e\n\u003cli\u003eHowever, how it reshapes their internal mechanisms remains poorly understood.\u003c/li\u003e\n\u003cli\u003eTo address this, we investigate how fine-tuning alters internal representations in LLMs, including attention patterns and layer-wise activations, and examine whether these changes are linked to task-relevant components identified by EAP (e.g., attention heads and logit-level activations) that drive task performance.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21113v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Fine-tuning has emerged as a widely adopted approach for adapting LLMs to a variety of downstream tasks\u003c/li\u003e\n\u003cli\u003eHowever, how it reshapes their internal mechanisms remains poorly understood\u003c/li\u003e\n\u003cli\u003eTo address this, we investigate how fine-tuning alters internal representations in LLMs, including attention patterns and layer-wise activations, and examine wh…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21139\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eTinyCeNN-LM: Quality-Gated Conversion of Pretrained Attention with CeNN-Inspired Cellular-Recurrent Layers\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21139v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Replacing attention in a pretrained language model is a compatibility problem: a plausible substitute may alter representations expected by later layers.\u003c/li\u003e\n\u003cli\u003eTinyCeNN-LM introduces a \\emph{quality-gated post-training conversion} framework using CeNN-inspired cellular-recurrent layers with bounded local processing, compact recurrent memory, routing, fusion, and accept-or-rollback validation.\u003c/li\u003e\n\u003cli\u003eThree implementations are studied: Integrated Memory, MemoryFusion, and PDelta3-GDN2-CLVR+Local32.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21139v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Replacing attention in a pretrained language model is a compatibility problem: a plausible substitute may alter representations expected by later laye…\u003c/li\u003e\n\u003cli\u003eTinyCeNN-LM introduces a \\emph{quality-gated post-training conversion} framework using CeNN-inspired cellular-recurrent layers with bounded local processing, co…\u003c/li\u003e\n\u003cli\u003eThree implementations are studied: Integrated Memory, MemoryFusion, and PDelta3-GDN2-CLVR+Local32\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21149\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eClinician-Grounded Quality Assurance for AI-Assisted Psychiatric Intake\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21149v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Before patients can use AI-assisted psychiatric intake systems, health systems need practical ways to routinely evaluate these tools against their clinical standards for quality assurance.\u003c/li\u003e\n\u003cli\u003eBecause clinicians may use different intake styles, evaluation for this task must (1) support comparison across interviewing approaches, (2) minimize clinician burden, and (3) measure clinically relevant performance for health systems deploying these technologies.\u003c/li\u003e\n\u003cli\u003eWe present a clinician-grounded evaluation platform built around a memory-augmented patient simulator for open-ended AI interviewing, InterviewPlayground.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21149v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Before patients can use AI-assisted psychiatric intake systems, health systems need practical ways to routinely evaluate these tools against their cli…\u003c/li\u003e\n\u003cli\u003eBecause clinicians may use different intake styles, evaluation for this task must (1) support comparison across interviewing approaches, (2) minimize clinician…\u003c/li\u003e\n\u003cli\u003eWe present a clinician-grounded evaluation platform built around a memory-augmented patient simulator for open-ended AI interviewing, InterviewPlayground\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21157\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eCan Agents Design Better Chips with a Higher Level Abstraction?\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21157v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Large Language Model (LLM) agents are increasingly being explored for chip design, but most existing approaches operate directly at RTL.\u003c/li\u003e\n\u003cli\u003eWe ask whether agents can design better chips by leveraging higher-level abstractions.\u003c/li\u003e\n\u003cli\u003eWe compare Direct RTL Design, Agent-based HLS Design, Post-Compiler HLS Refinement, and Post-HLS RTL Refinement, and combine Agent-based HLS Design with Post-HLS RTL Refinement as Agent-based HLS with RTL Refinement (AHRR).\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21157v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Large Language Model (LLM) agents are increasingly being explored for chip design, but most existing approaches operate directly at RTL\u003c/li\u003e\n\u003cli\u003eWe ask whether agents can design better chips by leveraging higher-level abstractions\u003c/li\u003e\n\u003cli\u003eWe compare Direct RTL Design, Agent-based HLS Design, Post-Compiler HLS Refinement, and Post-HLS RTL Refinement, and combine Agent-based HLS Design with Post-HL…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.21165\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eSpecOpt: Contact-Diff Reasoning for Agentic Molecule Optimization Toward Binding Specificity\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.21165v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Off-target protein binding is a major source of adverse effects for small-molecule drugs, yet most structure-based molecular design methods focus on generating selective compounds de novo rather than improving the selectivity of existing, well- characterized drugs.\u003c/li\u003e\n\u003cli\u003eWe introduce specificity optimization (SpecOpt), a molecular design task that seeks constrained structural modifications to an existing compound that increase its binding preference for an intended target over known off-targets while preserving its structural identity and drug-like properties.\u003c/li\u003e\n\u003cli\u003eTo enable systematic evaluation, we construct a ChEMBL-derived benchmark from compound-target interaction data, identifying intended targets through curated drug-mechanism annotations and off- targets through measured activities.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.21165v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Off-target protein binding is a major source of adverse effects for small-molecule drugs, yet most structure-based molecular design methods focus on g…\u003c/li\u003e\n\u003cli\u003eWe introduce specificity optimization (SpecOpt), a molecular design task that seeks constrained structural modifications to an existing compound that increase i…\u003c/li\u003e\n\u003cli\u003eTo enable systematic evaluation, we construct a ChEMBL-derived benchmark from compound-target interaction data, identifying intended targets through curated dru…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"arxiv-cscl-b_introsearch\"\u003e\n  ArXiv cs.CL (B_intro+search)\n  \u003ca class=\"heading-link\" href=\"#arxiv-cscl-b_introsearch\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20824\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eDo small language models know what they don\u0026rsquo;t know?\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20824v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: We explore whether entropy-based confidence signals can be leveraged to improve the accuracy of Small Language Models (SLMs) with fewer than 3 billion parameters, running entirely on consumer hardware.\u003c/li\u003e\n\u003cli\u003eWe evaluate seven distinct approaches, including token-level entropy early stopping, semantic entropy estimation, and uncertainty-aware routing to larger expert models, across 7 model pairs and 5 standard NLU benchmarks.\u003c/li\u003e\n\u003cli\u003eOur key finding is that token-level entropy is effectively blind in SLMs: in 91% of dataset-model combinations, mean token entropy is near zero regardless of answer correctness, rendering token-based confidence signals unusable at this scale.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20824v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: We explore whether entropy-based confidence signals can be leveraged to improve the accuracy of Small Language Models (SLMs) with fewer than 3 billion…\u003c/li\u003e\n\u003cli\u003eWe evaluate seven distinct approaches, including token-level entropy early stopping, semantic entropy estimation, and uncertainty-aware routing to larger expert…\u003c/li\u003e\n\u003cli\u003eOur key finding is that token-level entropy is effectively blind in SLMs: in 91% of dataset-model combinations, mean token entropy is near zero regardless of an…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20825\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eHERMES: Contrast-Aware Knowledge Graph Reasoning from Clinical Notes for Patient Outcome Prediction\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20825v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Clinical predictive models often rely on structured Electronic Health Record data, such as time-series and procedure codes.\u003c/li\u003e\n\u003cli\u003eWhile recent approaches have begun leveraging unstructured clinical notes, they typically encode them as flat sequences, which may lose explicit relational and temporal structure present in clinical narratives.\u003c/li\u003e\n\u003cli\u003eIn response, we propose HERMES, a graph-based framework that operates exclusively on clinical text while preserving clinical relationships.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20825v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Clinical predictive models often rely on structured Electronic Health Record data, such as time-series and procedure codes\u003c/li\u003e\n\u003cli\u003eWhile recent approaches have begun leveraging unstructured clinical notes, they typically encode them as flat sequences, which may lose explicit relational and…\u003c/li\u003e\n\u003cli\u003eIn response, we propose HERMES, a graph-based framework that operates exclusively on clinical text while preserving clinical relationships\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20826\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eTALON: A Temporally Aware Longitudinal Framework for Radiology Report Generation\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20826v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Current radiology report generation (RRG) models usually produce descriptive reports based on a single examination or only the most recent prior examination, limiting their ability to perform accurate and meaningful longitudinal comparisons and detect subtle interval changes.\u003c/li\u003e\n\u003cli\u003eAlthough recent approaches have begun to incorporate multiple prior examinations, they usually aggregate a fixed-length history without explicitly modeling the role-dependent relevance of each prior examination before fusion.\u003c/li\u003e\n\u003cli\u003eTo address this, we propose TALON, a Temporally Aware LONgitudinal RRG framework that adaptively integrates variable-length patient histories.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20826v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Current radiology report generation (RRG) models usually produce descriptive reports based on a single examination or only the most recent prior exami…\u003c/li\u003e\n\u003cli\u003eAlthough recent approaches have begun to incorporate multiple prior examinations, they usually aggregate a fixed-length history without explicitly modeling the…\u003c/li\u003e\n\u003cli\u003eTo address this, we propose TALON, a Temporally Aware LONgitudinal RRG framework that adaptively integrates variable-length patient histories\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20827\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eFrom Discharge Notes to Patient Understanding: Persona-Grounded, Open-Ended Simulation of LLMs as Discharge Educators\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20827v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Hospital discharge education is an interactive teaching task: a clinician adapts a discharge plan to a patient\u0026rsquo;s literacy, recall, and personality.\u003c/li\u003e\n\u003cli\u003eExisting LLM evaluations target static or artifact-generation tasks and do not measure patient understanding under open-ended dialogue.\u003c/li\u003e\n\u003cli\u003eWe introduce DischargeBench, a persona-grounded simulation in which a candidate LLM educator conducts a multi-turn session with a Virtual Patient, while an Education Monitor Agent regulates patient realism without modifying the educator, protecting the evaluation signal.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20827v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Hospital discharge education is an interactive teaching task: a clinician adapts a discharge plan to a patient\u0026rsquo;s literacy, recall, and personality\u003c/li\u003e\n\u003cli\u003eExisting LLM evaluations target static or artifact-generation tasks and do not measure patient understanding under open-ended dialogue\u003c/li\u003e\n\u003cli\u003eWe introduce DischargeBench, a persona-grounded simulation in which a candidate LLM educator conducts a multi-turn session with a Virtual Patient, while an Educ…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20828\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eBeyond WER: Entity and Disfluency Recall in Accented Conversational ASR\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20828v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: ASR systems optimised for Word Error Rate (WER) often miss named entities and filled pauses in accented conversational English, both critical for language-learning feedback.\u003c/li\u003e\n\u003cli\u003eWe present a three-stage pipeline for speakers from India, Indonesia, and Latin America: (1) heuristic SQL filters curating entity-rich training data at 2.8x the entity density of random sampling, (2) regional LoRA adapters fine-tuned on Qwen2.5-Omni-3B producing both verbatim and corrected transcripts in a single forward pass, and (3) a six-category error taxonomy validated by an LLM-based judge (83.8% agreement, 210 human-labelled samples).\u003c/li\u003e\n\u003cli\u003eThe pipeline achieves 80-85% entity recall (up from 53-55%), 76-86% filler recall (up from \u0026lt;5%), and 6-10% WER across 6k test utterances, outperforming Whisper and a commercial ASR on entity recall while matching a zero-shot 30B model with 10x fewer parameters.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20828v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: ASR systems optimised for Word Error Rate (WER) often miss named entities and filled pauses in accented conversational English, both critical for lang…\u003c/li\u003e\n\u003cli\u003eWe present a three-stage pipeline for speakers from India, Indonesia, and Latin America: (1) heuristic SQL filters curating entity-rich training data at 2.8x th…\u003c/li\u003e\n\u003cli\u003eThe pipeline achieves 80-85% entity recall (up from 53-55%), 76-86% filler recall (up from \u0026lt;5%), and 6-10% WER across 6k test utterances, outperforming Whisper…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20829\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eSAGE: Schema-Guided LLMs for Grant Review\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20829v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Grant reviewers must apply detailed criteria to application forms, budgets, and supporting documents while producing assessments that colleagues can inspect.\u003c/li\u003e\n\u003cli\u003eWe present SAGE, Schema-Guided Aspect-Based Grant Evaluation, a system that translates a grant rubric into structured checks and links its judgements to evidence from the application package.\u003c/li\u003e\n\u003cli\u003eWe evaluate SAGE in two stages on 35 nonprofit grant applications.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20829v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Grant reviewers must apply detailed criteria to application forms, budgets, and supporting documents while producing assessments that colleagues can i…\u003c/li\u003e\n\u003cli\u003eWe present SAGE, Schema-Guided Aspect-Based Grant Evaluation, a system that translates a grant rubric into structured checks and links its judgements to evidenc…\u003c/li\u003e\n\u003cli\u003eWe evaluate SAGE in two stages on 35 nonprofit grant applications\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20830\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eReviser: Revision-Capable Text Generation via Autoregressive Cursor Actions\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20830v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Revision-capable generation is appealing because it can insert or revise earlier content, but many non-autoregressive and edit-based approaches obtain this flexibility through repeated sequence-level computation.\u003c/li\u003e\n\u003cli\u003eWe propose Reviser, a decoder-only Transformer that generates a response as a sequence of cursor-relative actions on a mutable canvas.\u003c/li\u003e\n\u003cli\u003eAt each step, Reviser predicts exactly one action token: INSERT(token), MOVE($\\Delta$), or STOP, and is autoregressive over edit-history actions rather than final text order.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20830v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Revision-capable generation is appealing because it can insert or revise earlier content, but many non-autoregressive and edit-based approaches obtain…\u003c/li\u003e\n\u003cli\u003eWe propose Reviser, a decoder-only Transformer that generates a response as a sequence of cursor-relative actions on a mutable canvas\u003c/li\u003e\n\u003cli\u003eAt each step, Reviser predicts exactly one action token: INSERT(token), MOVE($\\Delta$), or STOP, and is autoregressive over edit-history actions rather than fin…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20831\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eRecursive Language Models Generalize Out of Domain\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20831v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: We study when limiting what a language model can see improves learning.\u003c/li\u003e\n\u003cli\u003eWe compare standard CoT, the more general learner that reads the full trace, with recursive language models, which restricts itself by solving each subtask in an isolated context.\u003c/li\u003e\n\u003cli\u003eIn-distribution, this generality comes for free: CoT can efficiently simulate the recursive rule, so the IID generalization guarantee changes only by a constant factor, and recursion does not offer much.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20831v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: We study when limiting what a language model can see improves learning\u003c/li\u003e\n\u003cli\u003eWe compare standard CoT, the more general learner that reads the full trace, with recursive language models, which restricts itself by solving each subtask in a…\u003c/li\u003e\n\u003cli\u003eIn-distribution, this generality comes for free: CoT can efficiently simulate the recursive rule, so the IID generalization guarantee changes only by a constant…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20832\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eTatBLiMP: A Benchmark of Linguistic Minimal Pairs for Tatar\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20832v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: We introduce TatBLiMP, the first benchmark of linguistic minimal pairs for Tatar (tt, ISO 639-3 tat), a Qypchaq Turkic language written in Cyrillic.\u003c/li\u003e\n\u003cli\u003eTo our knowledge it is the first grammaticality evaluation for Tatar language models of any kind, since even the 101-language MultiBLiMP does not include Tatar.\u003c/li\u003e\n\u003cli\u003eTatBLiMP covers 16 morphosyntactic phenomena in 1248 sentence pairs.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20832v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: We introduce TatBLiMP, the first benchmark of linguistic minimal pairs for Tatar (tt, ISO 639-3 tat), a Qypchaq Turkic language written in Cyrillic\u003c/li\u003e\n\u003cli\u003eTo our knowledge it is the first grammaticality evaluation for Tatar language models of any kind, since even the 101-language MultiBLiMP does not include Tatar\u003c/li\u003e\n\u003cli\u003eTatBLiMP covers 16 morphosyntactic phenomena in 1248 sentence pairs\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20833\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eTranssion\u0026rsquo;s Speaker-Attributed Multilingual ASR System for the MLC-SLM 2026 Challenge\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20833v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: This paper presents the Transsion Speech Team submission to Task 1 of the MLC-SLM 2026 Challenge, which focuses on speaker-attributed transcription for multilingual conversational speech.\u003c/li\u003e\n\u003cli\u003eWe propose a cascaded framework consisting of three components: a speaker diarization module, a long-form multilingual ASR module, and a speaker-transcription fusion module.\u003c/li\u003e\n\u003cli\u003eThe diarization module is built upon DiariZen and produces speaker-homogeneous segments through local speaker activity estimation and global speaker clustering.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20833v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: This paper presents the Transsion Speech Team submission to Task 1 of the MLC-SLM 2026 Challenge, which focuses on speaker-attributed transcription fo…\u003c/li\u003e\n\u003cli\u003eWe propose a cascaded framework consisting of three components: a speaker diarization module, a long-form multilingual ASR module, and a speaker-transcription f…\u003c/li\u003e\n\u003cli\u003eThe diarization module is built upon DiariZen and produces speaker-homogeneous segments through local speaker activity estimation and global speaker clustering\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003ch3 id=\"arxiv-cslg-b_introsearch\"\u003e\n  ArXiv cs.LG (B_intro+search)\n  \u003ca class=\"heading-link\" href=\"#arxiv-cslg-b_introsearch\"\u003e\n    \u003ci class=\"fa-solid fa-link\" aria-hidden=\"true\" title=\"链接到标题\"\u003e\u003c/i\u003e\n    \u003cspan class=\"sr-only\"\u003e链接到标题\u003c/span\u003e\n  \u003c/a\u003e\n\u003c/h3\u003e\n\u003cul\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20883\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eSparse Priors for Efficient Distribution Learning\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20883v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Despite the widespread use and success of generative AI techniques today, theoretical guarantees on learning a distribution supported in $d$ dimensions from $n$ samples degrade as $O(n^{-1/\\Theta(d)})$, though shown to be minimax optimal.\u003c/li\u003e\n\u003cli\u003eWe hypothesize that present bounds are too pessimistic because smoothness assumptions are not enough to capture the structure of distributions that often appear in real applications.\u003c/li\u003e\n\u003cli\u003eConsequently, we introduce the class of sparse priors and define the \u0026ldquo;Sparse Dimension\u0026rdquo; as a measure of sparsity of a prior over the space of all distributions.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20883v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Despite the widespread use and success of generative AI techniques today, theoretical guarantees on learning a distribution supported in $d$ dimension…\u003c/li\u003e\n\u003cli\u003eWe hypothesize that present bounds are too pessimistic because smoothness assumptions are not enough to capture the structure of distributions that often appear…\u003c/li\u003e\n\u003cli\u003eConsequently, we introduce the class of sparse priors and define the \u0026ldquo;Sparse Dimension\u0026rdquo; as a measure of sparsity of a prior over the space of all distributions\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20886\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eBI-Agent and BI-Bench: Towards Automating End-to-End Business Intelligence\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20886v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Business intelligence (BI) is a cornerstone of enterprise decision-making and is widely used by enterprise users in software such as Power BI and Tableau.\u003c/li\u003e\n\u003cli\u003eIn traditional BI workflows, users need to prepare data by (1) identifying relevant tables, (2) performing data transformations, and (3) building join relationships, before they can (4) answer their business questions.\u003c/li\u003e\n\u003cli\u003eThese steps can be complex and time-consuming, making BI challenging.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20886v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Business intelligence (BI) is a cornerstone of enterprise decision-making and is widely used by enterprise users in software such as Power BI and Tabl…\u003c/li\u003e\n\u003cli\u003eIn traditional BI workflows, users need to prepare data by (1) identifying relevant tables, (2) performing data transformations, and (3) building join relations…\u003c/li\u003e\n\u003cli\u003eThese steps can be complex and time-consuming, making BI challenging\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20888\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eElastic Threshold Attention: Learned Contextual Sparsity for Long-Context Decoding\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20888v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Massive KV caches can cause severe memory-bandwidth bottlenecks during long-context decoding.\u003c/li\u003e\n\u003cli\u003eSparse attention methods mitigate this via selective loading, but that comes at a cost: rigid heuristics drop necessary context, leading to quality degradation.\u003c/li\u003e\n\u003cli\u003eWe introduce \\textbf{Elastic Threshold Attention (ETA)}, an end-to-end trainable architecture that achieves hardware-accelerated decoding speed without sacrificing dense model quality.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20888v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Massive KV caches can cause severe memory-bandwidth bottlenecks during long-context decoding\u003c/li\u003e\n\u003cli\u003eSparse attention methods mitigate this via selective loading, but that comes at a cost: rigid heuristics drop necessary context, leading to quality degradation\u003c/li\u003e\n\u003cli\u003eWe introduce \\textbf{Elastic Threshold Attention (ETA)}, an end-to-end trainable architecture that achieves hardware-accelerated decoding speed without sacrific…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20904\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eBio-MF: Low-Latency and High-Fidelity EEG-to-fNIRS Cross-Modal Generation for Hybrid Motor-Imagery Brain\u0026ndash;Computer Interfaces\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20904v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Hybrid motor-imagery brain-computer interfaces (MI-BCIs) combining EEG and fNIRS can outperform EEG-only systems by exploiting complementary electrophysiological and hemodynamic information.\u003c/li\u003e\n\u003cli\u003eTo obtain such hybrid information when paired EEG-fNIRS acquisition is unavailable or inconvenient, recent studies have focused on EEG-to-fNIRS cross-modal generation.\u003c/li\u003e\n\u003cli\u003eHowever, existing methods still suffer from slow generation and often require pretraining, limiting their use in real-time MI-BCI scenarios.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20904v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Hybrid motor-imagery brain-computer interfaces (MI-BCIs) combining EEG and fNIRS can outperform EEG-only systems by exploiting complementary electroph…\u003c/li\u003e\n\u003cli\u003eTo obtain such hybrid information when paired EEG-fNIRS acquisition is unavailable or inconvenient, recent studies have focused on EEG-to-fNIRS cross-modal gene…\u003c/li\u003e\n\u003cli\u003eHowever, existing methods still suffer from slow generation and often require pretraining, limiting their use in real-time MI-BCI scenarios\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20906\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eContinuous Delayed-Memory Stochastic Gradient Descent and Continuous-Time Reinforcement Learning from History of Astrophysical Time Series Studies\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20906v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Quasars are luminous objects in the universe that exhibit stochastic brightness variations encoding information about the supermassive black holes powering them, and modeling these variations from ground-based survey data time series, known as light curves, is a statistical challenge.\u003c/li\u003e\n\u003cli\u003eThis paper reviews how stochastic differential equations (SDEs) have been adapted with neural network parameterizations to overcome this challenge in history.\u003c/li\u003e\n\u003cli\u003eWe create the Continuous-Delayed-Memory Stochastic Gradient Descent which depend on the past state of the discrete iteration process.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20906v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Quasars are luminous objects in the universe that exhibit stochastic brightness variations encoding information about the supermassive black holes pow…\u003c/li\u003e\n\u003cli\u003eThis paper reviews how stochastic differential equations (SDEs) have been adapted with neural network parameterizations to overcome this challenge in history\u003c/li\u003e\n\u003cli\u003eWe create the Continuous-Delayed-Memory Stochastic Gradient Descent which depend on the past state of the discrete iteration process\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20912\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eDo Quantum Models Scale Like LLMs?\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20912v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: In this work, we study the neural scaling laws of RydbergGPT, an autoregressive transformer model trained on qubit projective measurement data gathered from interacting Rydberg atom arrays.\u003c/li\u003e\n\u003cli\u003eThe quantum system is known to exhibit a finite-size remnant of a critical point as the laser detuning parameter is varied.\u003c/li\u003e\n\u003cli\u003eWe find that near the critical point the transformer loss as a function of training dataset size is well described by a power-law with a loss floor correction.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20912v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: In this work, we study the neural scaling laws of RydbergGPT, an autoregressive transformer model trained on qubit projective measurement data gathere…\u003c/li\u003e\n\u003cli\u003eThe quantum system is known to exhibit a finite-size remnant of a critical point as the laser detuning parameter is varied\u003c/li\u003e\n\u003cli\u003eWe find that near the critical point the transformer loss as a function of training dataset size is well described by a power-law with a loss floor correction\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20942\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eWhen AI Reviews Train AI Reviewers: Scientific-Judgment Collapse and Mitigation\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20942v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: Large language models (LLMs) increasingly participate in scientific evaluation, both as automated reviewers and as assistants to human reviewers.\u003c/li\u003e\n\u003cli\u003eAs model-generated reviews enter public data and future training corpora, AI peer review can become recursive: later reviewers learn from judgments produced by earlier models.\u003c/li\u003e\n\u003cli\u003eWe study one step of this feedback loop in a controlled setting.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20942v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: Large language models (LLMs) increasingly participate in scientific evaluation, both as automated reviewers and as assistants to human reviewers\u003c/li\u003e\n\u003cli\u003eAs model-generated reviews enter public data and future training corpora, AI peer review can become recursive: later reviewers learn from judgments produced by…\u003c/li\u003e\n\u003cli\u003eWe study one step of this feedback loop in a controlled setting\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20954\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eEfficient Bayes-Adaptive Reinforcement Learning with Temporal Logic Specifications\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20954v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: We present a novel end-to-end model-based Reinforcement Learning (RL) algorithm for efficient policy synthesis under given Linear Temporal Logic (LTL) specifications (e.g., safety or reachability) in unknown environments.\u003c/li\u003e\n\u003cli\u003eTo do so, a Limit-Deterministic B{\u0026quot;u}chi Automaton (LDBA) representation of the LTL task is synchronised with a Bayes-Adaptive Markov Decision Process (BAMDP) representation of the environment, which allows us to leverage an enhanced exploration-exploitation trade-off that is achieved via Bayesian RL, as opposed to traditional non-Bayesian approaches.\u003c/li\u003e\n\u003cli\u003eWe further propose a novel Bayes-Adaptive Monte-Carlo Planning (BAMCP) algorithm to allow for approximate Bayes-optimal strategy synthesis in the synchronised BAMDP construct.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20954v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: We present a novel end-to-end model-based Reinforcement Learning (RL) algorithm for efficient policy synthesis under given Linear Temporal Logic (LTL)…\u003c/li\u003e\n\u003cli\u003eTo do so, a Limit-Deterministic B{\u0026quot;u}chi Automaton (LDBA) representation of the LTL task is synchronised with a Bayes-Adaptive Markov Decision Process (BAMDP)…\u003c/li\u003e\n\u003cli\u003eWe further propose a novel Bayes-Adaptive Monte-Carlo Planning (BAMCP) algorithm to allow for approximate Bayes-optimal strategy synthesis in the synchronised B…\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20968\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eFrom Switching to Dynamic Regret: A Simple Reduction via Unbiased Random Sequences\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20968v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: In non-stationary online learning, dynamic regret has attracted increasing attention as a measure of how well an online learner performs against a time-varying comparator sequence.\u003c/li\u003e\n\u003cli\u003eDespite considerable advances, attaining optimal bounds for strongly convex and exp-concave losses often involves intricate analysis.\u003c/li\u003e\n\u003cli\u003eIn this paper, we present a \\textit{simple} framework that reduces dynamic regret minimization to switching regret minimization.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20968v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: In non-stationary online learning, dynamic regret has attracted increasing attention as a measure of how well an online learner performs against a tim…\u003c/li\u003e\n\u003cli\u003eDespite considerable advances, attaining optimal bounds for strongly convex and exp-concave losses often involves intricate analysis\u003c/li\u003e\n\u003cli\u003eIn this paper, we present a \\textit{simple} framework that reduces dynamic regret minimization to switching regret minimization\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003e\n\u003cp\u003e\u003cstrong\u003e\u003ca href=\"https://arxiv.org/abs/2609.20978\"  class=\"external-link\" target=\"_blank\" rel=\"noopener\"\u003eGenerative inversion for early ranking of competing geologic interpretations\u003c/a\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cul\u003e\n\u003cli\u003e发布时间:2026-09-21 12:00 北京时间\u003c/li\u003e\n\u003cli\u003e摘要:【待翻译】- arXiv:2609.20978v1 Announce Type: new.\n\u003cul\u003e\n\u003cli\u003eAbstract: High-consequence subsurface decisions are often made under severe data scarcity.\u003c/li\u003e\n\u003cli\u003eExperts may arrive at competing interpretations of the same subsurface system, yet early in a project there is rarely a practical way to determine which one is most realistic.\u003c/li\u003e\n\u003cli\u003eThis uncertainty can persist until several wells are drilled, often costing millions of dollars.\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003cli\u003eEN 要点:\n\u003cul\u003e\n\u003cli\u003earXiv:2609.20978v1 Announce Type: new\u003c/li\u003e\n\u003cli\u003eAbstract: High-consequence subsurface decisions are often made under severe data scarcity\u003c/li\u003e\n\u003cli\u003eExperts may arrive at competing interpretations of the same subsurface system, yet early in a project there is rarely a practical way to determine which one is…\u003c/li\u003e\n\u003cli\u003eThis uncertainty can persist until several wells are drilled, often costing millions of dollars\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n\u003c/li\u003e\n\u003c/ul\u003e\n",
  "wordCount": 5727,
  "readingTime": 27,
  "tableOfContents": "\u003cnav id=\"TableOfContents\"\u003e\n  \u003cul\u003e\n    \u003cli\u003e\u003ca href=\"#-本期-watch-list-深度导读\"\u003e📖 本期 Watch List 深度导读\u003c/a\u003e\u003c/li\u003e\n    \u003cli\u003e\u003ca href=\"#-x-平台-ai-热点快讯\"\u003e🌐 X 平台 AI 热点快讯\u003c/a\u003e\n      \u003cul\u003e\n        \u003cli\u003e\u003ca href=\"#话题-1zai-open-sources-zcode-after-secret-code-uploads\"\u003e话题 1:Z.ai Open-Sources ZCode After Secret Code Uploads\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#话题-2spacexai-engineer-ships-2500-prs-using-grok-bot-agent-teams\"\u003e话题 2:SpaceXAI Engineer Ships 2,500 PRs Using Grok Bot Agent Teams\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#话题-3ling-30-flash-vl-ai-transforms-product-photos-into-marketing-plans-and-prototypes\"\u003e话题 3:Ling-3.0-flash-VL AI Transforms Product Photos into Marketing Plans and Prototypes\u003c/a\u003e\u003c/li\u003e\n      \u003c/ul\u003e\n    \u003c/li\u003e\n    \u003cli\u003e\u003ca href=\"#-大佬观点influencer-insights\"\u003e💡 大佬观点(Influencer Insights)\u003c/a\u003e\u003c/li\u003e\n    \u003cli\u003e\u003ca href=\"#-附录今日-watch-list-更新源列表\"\u003e📚 附录:今日 Watch List 更新源列表\u003c/a\u003e\n      \u003cul\u003e\n        \u003cli\u003e\u003ca href=\"#all-in-podcast-a_full\"\u003eAll-In Podcast (A_full)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#stratechery-by-ben-thompson-a_full\"\u003eStratechery by Ben Thompson (A_full)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#openai-blog-a_full\"\u003eOpenAI Blog (A_full)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#arxiv-csai-b_introsearch\"\u003eArXiv cs.AI (B_intro+search)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#arxiv-cscl-b_introsearch\"\u003eArXiv cs.CL (B_intro+search)\u003c/a\u003e\u003c/li\u003e\n        \u003cli\u003e\u003ca href=\"#arxiv-cslg-b_introsearch\"\u003eArXiv cs.LG (B_intro+search)\u003c/a\u003e\u003c/li\u003e\n      \u003c/ul\u003e\n    \u003c/li\u003e\n  \u003c/ul\u003e\n\u003c/nav\u003e",
  "isDraft": false
}