What is Hackers' Pub?

Hackers' Pub is a place for software engineers to share their knowledge and experience with each other. It's also an ActivityPub-enabled social network, so you can follow your favorite hackers in the fediverse and get their latest posts in your feed.

동네에 무늬가 비슷한 고양이 세 마리가 있고 구글포토는 이들을 구분하는데 상당한 어려움을 겪고 있다. ...... 하지만 참을성을 가지고 계속 가르쳐주니 최근엔 제법 잘 구분하기 시작했다.

0
0
0
0

On the limits of LLMs (Large Language models) and LRMs (Large Reasoning Models). The TL;DR: "Our findings reveal fundamental limitations in current models: despite sophisticated self-reflection mechanisms, these models fail to develop generalizable reasoning capabilities beyond certain complexity thresholds." Meaning: accuracy collapse.

Interesting paper from Apple. ml-site.cdn-apple.com/papers/t

• We question the current evaluation paradigm of LRMs on established math benchmarks and
design a controlled experimental testbed by leveraging algorithmic puzzle environments that enable
controllable experimentation with respect to problem complexity.
• We show that state-of-the-art LRMs (e.g., o3-mini, DeepSeek-R1, Claude-3.7-Sonnet-Thinking)
still fail to develop generalizable problem-solving capabilities, with accuracy ultimately collapsing
to zero beyond certain complexities across different environments.
• We find that there exists a scaling limit in the LRMs’ reasoning effort with respect to problem
complexity, evidenced by the counterintuitive decreasing trend in the thinking tokens after a
complexity point.
• We question the current evaluation paradigm based on final accuracy and extend our evaluation
to intermediate solutions of thinking traces with the help of deterministic puzzle simulators. Our
analysis reveals that as problem complexity increases, correct solutions systematically emerge at
later positions in thinking compared to incorrect ones, providing quantitative insights into the
self-correction mechanisms within LRMs.
• We uncover surprising limitations in LRMs’ ability to perform exact computation, including their
failure to benefit from explicit algorithms and their inconsistent reasoning across puzzle types.
0
0
0
0

日本のソフトウェア開発者の皆様をHackers' Pubにご招待します。Hackers' Pubは、ActivityPubを実装するソフトウェア開発者のためのSNSであり、ブログプラットフォームです。MastodonとQiitaやZennをミックスしたような雰囲気です。短文(投稿)と長文(記事)の両方に対応しており、ActivityPub上では投稿はNote、記事はArticleとして表現されます。また、快適な技術ブログ執筆のために、TeX数式やGraphvizダイアグラムなど、さまざまなMarkdown拡張にも対応しています。ご興味のある方は、下記の招待リンクから先着25名までご登録いただけます。

https://hackers.pub/@hongminhee/invite/0197453c-95a9-7542-8c23-dc213ba07fb0

8
0
1
0
0
0
0
0
0
0
0
0

Hackers' Pub에 공유할 수 있는 초대 링크 기능을 추가했습니다. 설정 → 초대 페이지에서 초대 링크 생성이 가능하며, 초대 링크는 정원이나 유효 기간, 메시지를 정할 수 있게 되어 있습니다. 아무래도 DM으로 직접 초대 요청을 달라고 하면 쑥스러움이 많은 분들은 요청을 안 하시는 경우가 많은 것 같아서 만들게 되었습니다. 참고로, 초대 링크를 통해 가입한 사용자는 초대 링크를 생성한 사람과 자동으로 맞팔하게 됩니다.

초대 링크는 생성하는 화면초대 링크 페이지
9
1
0

"CDU, CSU und SPD lassen alle Hemmungen fallen. Sie planen auf gleich drei Ebenen: Telekommunikations-, Autokennzeichen- und Biometriedaten. Die alten und neuen riesigen Datenhalden sollen mit „automatisierter Datenrecherche und -analyse“ durchkämmt werden dürfen. Staatliches Hacken soll ebenfalls ausgeweitet werden.

Der Koalitionsvertrag, den die schwarz-schwarz-rote Regierung abschließen will, strotzt so vor Überwachungsvorhaben, dass jeder Einzelne betroffen sein wird. Ob man im Netz kommuniziert, Auto fährt oder Fotos mit Gesichtern ins Netz stellt: All das soll massenhaft aufgezeichnet und bei Bedarf ausgewertet werden.

Und nicht nur die massenhafte Überwachung und Ausweitung der Nutzung von Staatstrojanern soll kommen, vielmehr wollen Christ- und Sozialdemokraten den Paradigmenwechsel: Informationelle Selbstbestimmung soll auf den Scheiterhaufen, Datennutzung und der ganze „KI“-Quatsch sollen Priorität bekommen."

ccc.de/de/updates/2025/ueberwa

0
0
0
0

日本のソフトウェア開発者の皆様をHackers' Pubにご招待します。Hackers' Pubは、ActivityPubを実装するソフトウェア開発者のためのSNSであり、ブログプラットフォームです。MastodonとQiitaやZennをミックスしたような雰囲気です。短文(投稿)と長文(記事)の両方に対応しており、ActivityPub上では投稿はNote、記事はArticleとして表現されます。また、快適な技術ブログ執筆のために、TeX数式やGraphvizダイアグラムなど、さまざまなMarkdown拡張にも対応しています。ご興味のある方は、下記の招待リンクから先着25名までご登録いただけます。

https://hackers.pub/@hongminhee/invite/0197453c-95a9-7542-8c23-dc213ba07fb0

8
0
1
0
0
0
0
0

【上海高考作文出处作者胡晓明:考生可能想骂我,但可写还蛮多】6月7日上午11:30, 出炉。今年上海的高考作文题由“专”“转”“传”三个字而来,提及有学者用“专”“转”“传”概括三类文章,让不少网友直呼很难。上海文化学术圈很快“破案”,这位学者就是华东师范大学终身教授、上海写作学会会长胡晓明,他在一篇发于“文汇笔会”的《胡晓明:古人说“文章九命”,太悲观了,我归为三个字:专、转、传》里谈及“专、转、传”。“今年上海高考作文确实来自我这篇文章。”6月7日中午,澎湃新闻记者联系到胡晓明,他在电话那头说:“可能很多同学会骂我,觉得一看到这个题目会有点懵,因为他们不太适合这种很中国式的强调汉字的思维方式。但实际上静下来仔细想一想,可以写的东西还是蛮多的。”
对于“文章进高考”,他回应了三点。首先,“专、转、传”三个字是辩证关系,今年的高考作文题依然延续了哲学思辨,“要么就是躺在图书馆里的高文典章,要么就是流量至上的文字垃圾,在两个极端都不对的,中间一定有很多可以去思辨的。特别我们今天遇到了一个前所未有的传播时代,这个时代给了我们福分,也给了我们挑战,因为新的传播方式带来了很多机遇和问题。”第二,他认为这个题目引导考生参与当代文化生活与思想反思。“我想只要不是死读书的学生,不是只在课文里打转转的学生,他一定有很多鲜活的例子,有很多可以表达的的东西。”最后,他认为这道题也可以让更多考生尊重中国古典语文的思维方式,而不是西方的概念范畴。他说,中国的古典语文思维就是汉字文字思维,它背后有很多东西可以让学生们去深入学习和思索。更多详细内容请查看原文>> :sys_link: 网页链接 3g.k.sohu.com/t/n896759771?ser

:icon_weibo: weibo.com/5890672121/PvrbpDiUf

0
0
0
0
0
0
0
0
0
0
0

My partner of 13 years has turned into a serial moaner, so much so I have started creating a moan diary for my own amusement logging them. 18 moans yesterday about money, tiredness and "all these biscuits look the fucking same". Can't wait to see how many it is today, exciting.

0
0

Mathematics Teaching 296 now available online atm.org.uk/Mathematics-Teachin

Design and cover photographs by me

Four free articles for non-members:

Everyone can think mathematically by Tom Francome
Tom Francome explores ways of developing the mathematical thinking of all students, including low attainers.
atm.org.uk/write/MediaUploads/

Book review Learning with AI by Ian Benson
Ian Benson reviews 'Learning with AI' by Joan Monahan Watson published by Johns Hopkins University Press (296 pages, $24.95)
atm.org.uk/write/MediaUploads/

Book review Breaking images by Pete Wright
Pete Wright reviews ‘Breaking images: Iconoclastic analyses of mathematics and its education’, edited by Brian Greer, David Kollosche, and Ole Skovsmose.
atm.org.uk/write/MediaUploads/

Ole Skovsmose—the man who put the critique in critical mathematics education by Peter Gates
Peter Gates has collated this obituary for Ole Skovmose.
atm.org.uk/write/MediaUploads/

Cover of Mathematics Teaching 296, June 2025. The cover is mainly white and is dominated by four photographs of people at the Joint Conference of Mathematical Subject Associations. Top left: an audience applauds a speaker that is outside the picture. Top right: a smiling woman holds an octahedron made of yellow balloons and another woman looks on. Bottom left: people seated at tables look at a woman holding a large piece of paper with writing and drawing on it. Bottom right: a smiling man wears a hat made of red, yellow and purple balloons.
0
0
4
0
0
0
0
0
0
0
0

No, Von der Leyen. Healthcare won't be improved by .
On the contrary, AI is destroying the future of healthcare. Its engines are stealing researcher's work and atrophying the brains of future health professionals.
Medicine is a social science that needs critical thinking, empathy and compassion. All very distant concepts from those powerful shortsighted technocrats ruling the European Commission, the only beneficiaries of those €200 billion.

Commission President Ursula von der Leyen said:

“AI will improve our healthcare, spur our research and innovation and boost our competitiveness. We want AI to be a force for good and for growth. We are doing this through our own European approach – based on openness, cooperation and excellent talent. But our approach still needs to be supercharged. This is why, together with our Member States and with our partners, we will mobilise unprecedented capital through InvestAI for European AI gigafactories. This unique public-private partnership, akin to a CERN for AI, will enable all our scientists and companies – not just the biggest - to develop the most advanced very large models needed to make Europe an AI continent.”
0
0
0
0
0