DECOHACK

Hacker News Daily · No.078

HN 每日热榜 2026-09-16

Hacker News 当天(UTC)发布的帖子按分数排出前 30,标题译成中文,另选 Show HN 和 Ask HN 各 5 条。

上榜帖子
30
总分数
10,827
总评论
6,452
最高分
970

热门前 30

  1. 01
    Nvidia 宣布在 Rust 中进行原生 GPU 编程

    Nvidia announces native GPU programming in Rustdeveloper.nvidia.com

    970404
  2. 02
    训练一个 40 亿参数的模型,生成比 Postgres 快 81% 的查询计划

    Training a 4B model to produce 81% faster query plans than Postgresrohanbansal.com

    702144
  3. 03
    欧盟首席为加拿大成为“准成员”打开大门

    EU chief opens door for Canada to become 'associate member'bbc.com

    693905
  4. 04
    小编程技巧

    Small programming trickswill-keleher.com

    681294
  5. 05
    Mistral X Mozilla:私密的多语言 AI 浏览

    Mistral X Mozilla: Private, Multilingual AI Browsingmistral.ai

    591207
  6. 06
    黑客侵入了一台 Flock 摄像头

    Hackers Got Inside a Flock Camerawired.com

    586270
  7. 07
    小米 Mimo 2.6 实时后训练仪表板

    Xiaomi Mimo 2.6 live post-training dashboardmimo.xiaomi.com

    563155
  8. 08
    Apple 参考图像:一种新的验证摄影方法

    Apple Reference Image: A New Approach for Verified Photographysecurity.apple.com

    539352
  9. 09
    蜡电机

    Wax motoren.wikipedia.org

    525107
  10. 10
    Google Play 应用审核流程现在通常需要超过一周的时间

    The Google Play app review process now regularly takes longer than a weekgultsch.social

    374352
  11. 11
    备份并不简单

    Backups Aren't Simplefilipovski.net

    359225
  12. 12
    PS5 Linux 负责人辞职:"一群不懂的菜鸟在使用 LLM"

    PS5 Linux lead quits: "a bunch of noobs using LLMs" that "they don't understand"frvr.com

    344227
  13. 13
    原版 Sony PlayStation 2 安全芯片在 26 年后被完全破解

    Original Sony PlayStation 2 security chip 'broken wide open' after 26 yearstomshardware.com

    28591
  14. 14
    Salesforce 全球故障

    Salesforce Global Outagestatus.salesforce.com

    280184
  15. 15
    在 LLM 时代学习编程

    Learning Programming in an Age of LLMsblog.ploeh.dk

    263195
  16. 16
    澳大利亚表示可能会效仿加拿大与欧盟建立更深层次的联系

    Australia says it could follow Canada in forging deeper ties with EUindependent.co.uk

    254216
  17. 17
    突破三元 LLM 的 1.58 位障碍

    Breaking the 1.58-bit Barrier for Ternary LLMsarxiv.org

    24541
  18. 18
    关于“模型福利”的警告

    A warning about 'model welfare'mustafa-suleyman.ai

    242701
  19. 19
    Claude Cowork 和聊天现在合并为一个 Claude

    Claude Cowork and chat are now one Claudeclaude.com

    234225
  20. 20
    询问作者关于他们自己论文的事

    Asking authors about their own papersmedium.com

    234119
  21. 21
    HarnessTax:Harness 对编码代理的重要性有多大?

    HarnessTax: How Much Does the Harness Matter for Coding Agents?harnesstax.github.io

    23399
  22. 22
    Dream-RSI:通过进化世界实现递归自我改进

    Dream-RSI: Recursive Self-Improvement through Evolving Worldsarxiv.org

    21353
  23. 23
    OpenSpec – 一个轻量且可配置的 AI 规范框架

    OpenSpec – A lightweight and configurable AI spec frameworkopenspec.dev

    204100
  24. 24
    DeepMind 研究所

    The DeepMind Instituteinstitute.deepmind.com

    18676
  25. 25
    美联储加息,通胀担忧推高债券收益率

    Fed hikes rates as inflation worries push up bond yieldsreuters.com

    184250
  26. 26
    DeepSeek v4.1 Flash 现在是我们最好的黑客模型

    DeepSeek v4.1 Flash Is Now Our Best Hacking Modelenclave.ai

    17767
  27. 27
    向量化和性能可移植的快速排序(2022)

    Vectorized and performance-portable Quicksort (2022)opensource.googleblog.com

    17431
  28. 28
    逆向工程的 Jev 类模型

    Reverse-engineered Jev-like modelgithub.com

    16924
  29. 29
    在线赌博市场扩展后,赌博障碍的急诊就诊人数翻倍

    ER visits for gambling disorders doubled after expanded online gambling markettemertymedicine.utoronto.ca

    164157
  30. 30
    OpenAI 扩大 ChatGPT 广告,推出赞助代理

    OpenAI expands ChatGPT ads with Sponsored Agentsopenai.com

    159181

Show HN · 作品展示

Ask HN · 社区提问