CK's Curated Feed

• 📄 The Philosophy Book Dk

  6/8/2025, 10:11:47 PM — Exploratory-pdfs(local file)

  .PDF file (24.3 MB) from Exploratory-pdfs - Modified: 2025-06-08

• Chapter 1 of Ansi Common Lisp

  6/18/2025, 2:28:05 AM — paul_graham

• How to Start Google

  3/1/2024, 12:00:00 AM — paul_graham

• “大尺度變化的嘗試”， 記錄一下夢。

  10/18/2009, 9:36:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>昨天晚上做夢，我去一家房地產經紀公司當經紀人；夢有許多細節：如辦公室擺設、門店經理、夢的過程中的其它情景和模糊矛盾潛意識。<br/> 醒來之後，我想，去當一個月的經紀人也許很有意思，帶人看房、取得業主信任、這些都是我性格中很不擅長的事情， 或許完全不一樣的經歷能改變很多，而人生苦短，應該做更多的嘗試和改變。<br/> 當然，做這個情景夢，自我瞭解有其他緣由。<br/> 最近看《少有人走的路》，按作者的說法，潛意識裏我有改變自己的動力而在正常意識中不曾...

• Researchers and Founders

  6/19/2020, 5:39:12 PM — sam+altman

  I spent many years working with founders and now I work with researchers. Although there are always individual exceptions, on average it’s surprising to me how different the best people in these groups are (including in some qualities that I had assumed were present in great peop…

• Does AI make us stupider?

  6/25/2025, 5:04:48 AM — marginal_revolution

  That is the topic of my latest Free Press column , responding to a recent study out of MIT.  Here is one excerpt: To see how lopsided their approach is, consider a simple parable. It took me a lot of “cognitive load”—a key measure used in their paper—to memorize all those stat…

• Are cultural products getting longer?

  6/25/2025, 6:12:26 AM — marginal_revolution

  Ted Gioia argues that cultural products are getting longer : Some video creators have already figured this out. That’s why the number of videos longer than 20 minutes uploaded on YouTube  grew from 1.3 million to 8.5 million in just two years … Songs are also getting longer.…

• 📄 An Autonomous Ai Agent For Universal Behavior Analysis

  6/4/2025, 2:15:27 PM — AI papers(local file)

  .PDF file (10.8 MB) from AI papers - Modified: 2025-06-04

• Summary of John Halstead's Book-Length Report on Existential Risks From Climate Change

  6/25/2025, 3:14:44 PM — lesswrong

  Published on June 25, 2025 3:14 PM GMT 1 Introduction (Crossposted from my blog --formatting is better there).  Very large numbers of people seem to think that climate change is likely to end the world. Biden and Harris both called it an existential threat. AOC warned a few…

• CS 194/294-280 (Advanced LLM Agents) - Lecture 12, Dawn Song

  5/3/2025, 7:09:52 PM — yt_deep_dives

• I Tested LLM Agents on Simple Safety Rules. They Failed in Surprising and Informative Ways.

  6/25/2025, 9:39:08 PM — lesswrong

  Published on June 25, 2025 9:39 PM GMT TL;DR:  I developed a simple, open-source benchmark to test if LLM agents follow high-level safety principles when they conflict with a given task accepted at ICML 2025 Technical AI Governance Workshop (see  paper ). My pilot study on six …

• Rasheed Griffith on the economics and aesthetics of Asunción

  6/23/2025, 5:50:08 AM — marginal_revolution

  Yet, on my first visit to Asunción last week none of that was on my mind. What was striking was the total absence of any aesthetic coherence of the city. There are some economic reasons for this: Going back to the middle class consumption point. If only around 300,000 Paragua…

• Please Fund More Science

  3/30/2020, 5:46:36 PM — sam+altman

  Experts on the COVID-19 pandemic seem to think there are three ways out—that is, for life, health, and the economy to return roughly to normal.  Either we get a vaccine good enough that R0 for the world goes below 1, a good enough treatment that people no longer need to be afraid…

• 【生成式AI時代下的機器學習(2025)】第四講：Transformer 的時代要結束了嗎？介紹 Transformer 的競爭者們

  3/22/2025, 6:22:58 PM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1Xw59SuuOBwF9ea71hJ_8sdoWW2Jl9Dk6/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 25:10 跟訓練類神經網路有關的影片連結 Backpropagation: https://youtu.be/ibJpTrp5mcE Computational Graph: https://youtu.be/-yhm3WdGFok 32:03 應該是從輸入 x 到輸出…

• 正確的決策

  9/28/2024, 10:37:21 PM — sheracaolity

  公司轉眼間已經三歲了，我發現相較於創業第一年，當前最大的挑戰在於正確的決策不再像以前那麼顯而易見。創業早期雖然壓力大，但是因為用戶少、資源少，決策邏輯其實非常單純，只要把全部的注意力放在做產品、和用戶對話、讓營收成長上，並忽略其他噪音和會讓人分心的事物就行了。然而如今公司的用戶和資源都比當時多了許多，即便我把全部的注意力放在產品、用戶和營收上，也常常發現很多決策並不好做。用戶的需求很重要，但隨著用戶規模的成長，新用戶和老用戶在意的產品面向常常很不一樣；產品的開發速度很重要，但隨著團隊人數的擴張，新進員工和資深員工適合的開發流程常常很不一樣；營收的成長很重…

• 📄 Features Fusion Or Not Harnessing Multiple Pathological Foundation Models Using Meta Encoder For Downstream Tasks Fine Tuning

  6/19/2025, 7:51:43 PM — Research Papers(local file)

  .PDF file (5.9 MB) from Research Papers - Modified: 2025-06-19

• The Best Essay

  3/1/2024, 12:00:00 AM — paul_graham

• 📄 Predicting Hospital Readmission Via Cost Sensitive Deep Learning

  11/28/2024, 1:51:00 AM — Research Papers(local file)

  .PDF file (1.9 MB) from Research Papers - Modified: 2024-11-28

• What You (Want to)* Want

  11/1/2022, 12:00:00 AM — paul_graham

• CS 194/294-280 (Advanced LLM Agents) - Lecture 1, Xinyun Chen

  1/28/2025, 2:20:22 PM — yt_deep_dives

• Monday assorted links

  6/23/2025, 4:06:52 PM — marginal_revolution

  1. On the new Xi Zhongxun biography . 2. “ This month, four Finnish cities are offering six euros per litre for dead Spanish slugs. ” 3. “ There are new hints that  the fabric of space-time may be made of “memory cells ” that record the whole history of the universe. ” Specu…

• Stanford CS25: V2 I Robotics and Imitation Learning

  5/23/2023, 6:00:35 PM — yt_ai_fundamentals

  February 7, 2023 Robotics and Imitation Learning Ted Xiao In this speaker series, we examine the details of how transformers work, and dive deep into the different kinds of transformers and how they're applied in different fields. We do this by inviting people at the forefront o…

• 📄 Dwarkesh Patel The Scaling Era An Oral History Of Ai, 2019–2025 (2025, Stripe Press) Libgen.li

  6/25/2025, 11:16:53 AM — Books(local file)

  .EPUB file (2.4 MB) from Books - Modified: 2025-06-25

• Good Writing

  5/1/2025, 12:00:00 AM — paul_graham

• CS 194/294-280 (Advanced LLM Agents) - Lecture 4, Hanna Hajishirzi

  3/3/2025, 7:58:13 PM — yt_deep_dives

• Superlinear Returns

  10/1/2023, 12:00:00 AM — paul_graham

• Reflections

  1/6/2025, 1:37:29 AM — sam+altman

  The second birthday of ChatGPT was only a little over a month ago, and now we have transitioned into the next paradigm of models that can do complex reasoning. New years get people in a reflective mood, and I wanted to share some personal thoughts about how it has gone so far, an…

• Stanford CS25: V2 I Strategic Games

  5/22/2023, 6:00:23 PM — yt_ai_fundamentals

  January 31, 2023 Strategic Games Noam Brown In this speaker series, we examine the details of how transformers work, and dive deep into the different kinds of transformers and how they're applied in different fields. We do this by inviting people at the forefront of transformers…

• The Virus

  3/7/2020, 6:58:55 PM — sam+altman

  Although I still hope things will go differently, the experts I’ve spoken to think we are likely to face a global tragedy—hundreds of thousands of deaths from Covid-19. I hope that society views this as a warning for the future.  Covid-19 is bad, but only a warm-up.  I think it’s…

• 📄 Lora Vs Full Fine Tuning An Illusion Of Equivalence

  11/14/2024, 4:01:12 PM — AI papers(local file)

  .PDF file (8.1 MB) from AI papers - Modified: 2024-11-14

• Prover-Estimator Debate: A New Scalable Oversight Protocol

  6/17/2025, 1:53:04 PM — lesswrong

  Published on June 17, 2025 1:53 PM GMT Linkpost to arXiv: https://arxiv.org/abs/2506.13609 . Summary:  We present a scalable oversight protocol where honesty is incentivized at equilibrium. Prior debate protocols allowed a dishonest AI to force an honest AI opponent to solve a c…

• Chantilly destination achieved, the Limbourg brothers are amazing

  6/23/2025, 4:03:09 AM — marginal_revolution

  Commissioned by the Duc de Berry, the enormously wealthy brother of King Charles V of France,   this exquisite Book of Hours was begun by the Limbourg brothers, a trio of Netherlandish miniature painters, in around 1411. The Duc and the Limbourgs died in 1416. The manuscript was …

• 📄 Inferring Super Resolution Tissue Architecture By Integrating Spatial Transcriptomics With Histology

  2/7/2025, 4:56:52 PM — Research Papers(local file)

  .PDF file (27.1 MB) from Research Papers - Modified: 2025-02-07

• 📄 Adler, Mortimer J. Doren, Charles Van How To Read A Book The Classic Guide To Intelligent Reading (1972, Simon & Schuster, Inc) Libgen.li

  6/25/2025, 12:17:42 AM — Books(local file)

  .EPUB file (0.4 MB) from Books - Modified: 2025-06-25

• The Shape of the Essay Field

  6/1/2025, 12:00:00 AM — paul_graham

• Melatonin Self-Experiment Results

  6/25/2025, 3:58:10 PM — lesswrong

  Published on June 25, 2025 3:58 PM GMT Throughout the first half of 2025 I did a blinded experiment to see how low to moderate melatonin intake affects me. In this post I summarize my findings. Tl;dr : In my blinded self-experiment on melatonin, conducted over n = 60 days and ana…

• 📄 Cal Newport Deep Work Rules For Focused Success In A Distracted World Grand Central (2016)

  6/8/2025, 9:39:41 PM — Books(local file)

  .EPUB file (0.8 MB) from Books - Modified: 2025-06-08

• Hard Startups

  2/26/2020, 6:28:47 PM — sam+altman

  The most counterintuitive secret about startups is that it’s often easier to succeed with a hard startup than an easy one.  A hard startup requires a lot more money, time, coordination, or technological development than most startups. A good hard startup is one that will be valu…

• London Relocation: Taipei vs London

  4/30/2025, 2:23:11 PM — sheracaolity

  在過去十年，我有大約七年的時間住在台北、一年半的時間住在倫敦，這二個城市對我來說都已經算是熟悉的家了。以下是我對在這二個城市居住在不同面向上的評價： 飲食：台北完勝 餐廳：台北完勝。 在價格上，倫敦的普通餐廳隨便一餐都要二十磅左右，大約是台北的 2-3 倍。換句話說，若想在倫敦達到和台北相同的外食頻率，那麼稅後收入至少需要達到台北的二倍。 撇開價格不談，如果你喜歡東亞料理（中、日、韓、台），我覺得以台灣人喜歡的口味來說，台北餐廳的食物品質絕對是完勝倫敦的。這裡的完勝是指在東亞料理這個類別中，台灣當地的中價位餐廳很多時候是比倫敦當地的高價位餐廳還要好吃的，…

• London Relocation: Rent

  5/28/2024, 9:51:01 PM — sheracaolity

  上一篇文章 分享了申請英國簽證的心得，這篇文章想分享在英國租房子的心路歷程。 雖然去年十一月就拿到簽證了，但因為倫敦冬天非常冷、再加上想先在台灣過完年，所以我們直到今年三月才搬來倫敦。在來倫敦之前，我便已聽聞這邊的租屋市場相當競爭，所以事先和夥伴訂了一個月的飯店、為找房預留比較充裕的緩衝時間。我們的找房流程如下： 上網查詢其他人對倫敦不同區域的評價，從中篩選出所有我們覺得可以考慮居住的區域，並列出位在這些區域的地鐵站。在列了快二十個地鐵站後，我們用了整整二天的時間搭車去每一站逛半小時，感受整體氛圍、治安和生活機能，然後給予每一站一個評分。在所有地鐵站都逛…

• London Relocation: Global Talent Visa

  4/30/2024, 12:49:35 PM — sheracaolity

  前陣子做了一些公司的長期規劃後，決定和一些夥伴搬到倫敦工作，所以在半年前申請了英國的工作簽證，並在今年三月正式搬到英國。截至目前已經在倫敦住二個月了，趁現在記憶猶新，最近打算把一些過程和想法記錄下來。第一篇文就先分享一下我對英國簽證的看法。 我申請的 Global Talent Visa (Digital Technology) 是個對科技從業者相當友善的英國簽證，時效是 3-5 年（看你申請哪條路線），在英國時間待夠就可以轉 ILR（類似英國的綠卡），性質有點像美國的 O1 簽證（不需要企業贊助），但是申請的成本、門檻和準備時間比 O1 低很多，可以不…

• 📄 Dinov2 Learning Robust Visual Features Without Supervision

  1/2/2025, 2:03:41 AM — AI papers(local file)

  .PDF file (7.7 MB) from AI papers - Modified: 2025-01-02

• 【生成式AI時代下的機器學習(2025)】第六講：生成式人工智慧的後訓練(Post-Training)與遺忘問題

  4/19/2025, 5:49:12 PM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/129jFUXCPekD2Qoaq6H0Ntasm0GUiK8_X/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 33:00 1:14:15 我怕口頭上沒有說清楚，所以再用文字說明。我認為未來的趨勢是，多數語言模型仍需要具備自己獨特的能力（例如：特別擅長寫程式）。一個大型計劃可能需要由多個具備不同專長的語言模型組成團隊，共同完成，就像人類社會中的分工合作一樣。但要注意，所謂「特別擅長寫…

• 📄 Additive Mil Intrinsically Interpretable Multiple Instance Learning For Pathology

  11/11/2024, 4:13:19 PM — Research Papers(local file)

  .PDF file (22.4 MB) from Research Papers - Modified: 2024-11-11

• 📄 A Vision–language Foundation Model For Precision Oncology

  2/7/2025, 11:40:52 PM — Research Papers(local file)

  .PDF file (28.1 MB) from Research Papers - Modified: 2025-02-07

• CS 194/294-280 (Advanced LLM Agents) - Lecture 2, Jason Weston

  2/10/2025, 5:15:21 PM — yt_deep_dives

• [Linkpost] The lethal trifecta for AI agents: private data, untrusted content, and external communication

  6/17/2025, 4:09:08 PM — lesswrong

  Published on June 17, 2025 4:09 PM GMT TL;DR from the post by Simon Willison: The lethal trifecta of capabilities is: Access to your private data —one of the most common purposes of tools in the first place! Exposure to untrusted content —any mechanism by which text (or images)…

• Chapter 2 of Ansi Common Lisp

  6/18/2025, 2:28:05 AM — paul_graham

• 📄 Morris Chang 張忠謀 張忠謀自傳：下冊 一九六四──二○一八 (2024, 天下文化) Libgen.li

  6/25/2025, 12:26:16 AM — Books(local file)

  .EPUB file (23.0 MB) from Books - Modified: 2025-06-25

• 📄 Byrne Hobart Tobias Huber Boom Bubbles And The End Of Stagnation (2024, Stripe Press) Libgen.li

  6/10/2025, 12:51:22 AM — Books(local file)

  .EPUB file (1.1 MB) from Books - Modified: 2025-06-10

• CS 194/294-280 (Advanced LLM Agents) - Lecture 11, Swarat Chaudhuri

  4/28/2025, 6:17:11 PM — yt_deep_dives

• Heptabase November 2024 AMA

  11/29/2024, 10:03:44 AM — sheracaolity

  I hosted my second AMA in  Heptabase's Discord community this week. This post includes my responses to the top upvoted questions, which cover the following topics: Heptabase's plans for feature direction in 2025-Q1. Heptabase's plans for AI integration. Heptabase's plans for spa…

• 📄 A Pathology Foundation Model For Cancer Diagnosis And Prognosis Prediction

  11/27/2024, 11:19:12 PM — Research Papers(local file)

  .PDF file (37.0 MB) from Research Papers - Modified: 2024-11-27

• 有機會得到幫助和幫助別人都是很幸運的事

  9/19/2009, 8:39:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>續上篇打官司的日誌入題：除了租賃關係，借貸關係也是從步入社會，乃至步入大學就開始的社會關係。以不同心態對待這個關係很能體現對生活的態度。<br/> 有些人遇到借錢馬上想能不能還、自己有沒欠人情要借，其實對於很多金額不大的情況，爲什麼一定要想着還呢，一般需要借錢的情況一般是當事人因爲運氣不濟、個人迷失、挑戰失敗等導致的生活低潮，大多是很需要幫助的時候，報着或多或少盡點支持的心態多好。自己少一趟旅遊，晚一點買房，少上飯館，晚點換酷的電子設備等等。哪怕借款...

• 【生成式AI時代下的機器學習(2025)】第十講：人工智慧的微創手術 — 淺談 Model Editing

  5/4/2025, 10:36:19 AM — yt_ml_course

  本課程所講的 Model Editing，指的就是 Knowledge Editing，更多相關技術請見 Easy Edit 系列工具：https://github.com/zjunlp/EasyEdit 以及綜述論文：https://zjunlp.github.io/project/KnowEdit/ 字幕由網友 Xyfer 熱心提供 投影片連結：https://docs.google.com/presentation/d/10-o047tqJzpE4x77ga4eUAVNF3bqB663/edit?usp=sharing&ouid=1150460…

• AISN #57: The RAISE Act

  6/17/2025, 6:02:55 PM — lesswrong

  Published on June 17, 2025 6:02 PM GMT Welcome to the AI Safety Newsletter by the Center for AI Safety . We discuss developments in AI and AI safety. No technical background required. In this edition: The New York Legislature passes an act regulating frontier AI—but it may not b…

• 【生成式AI時代下的機器學習(2025)】第十一講：今天你想為 Foundation Model 裝備哪些 Task Vector？淺談神奇的 Model Merging 技術

  5/16/2025, 6:24:53 PM — yt_ml_course

  字幕由網友 Xyfer 熱心提供！ 投影片連結：https://docs.google.com/presentation/d/1PUEnnpaxuq8dtm_Z4OnYwYjUAz578w74/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 在課程中舉了許多 Model Merging 成功的例子，且這些例子在論文中看起來具有高度的泛用性。然而我想提醒一下同學，Model Merging 並不總是能成功，成效往往會受到任務性質的影響。有些任務較容易融合成功，例如經驗上在 1…

• 📄 Naval

  5/20/2024, 4:09:28 AM — Exploratory-pdfs(local file)

  .PDF file (1.8 MB) from Exploratory-pdfs - Modified: 2024-05-20

• 【生成式AI時代下的機器學習(2025)】助教課：利用多張GPU訓練大型語言模型—從零開始介紹DeepSpeed、Liger Kernel、Flash Attention及Quantization

  3/29/2025, 3:43:39 PM — yt_ml_course

  投影片連結（打開Excalidraw匯入即可）：https://drive.google.com/file/d/1pKgY4EuyXh7SuTbI_r7it1CK5DCFUKYZ/view?usp=sharing 歡迎同學到slido上匿名發問（連結會在4/4失效）：https://app.sli.do/event/o69HrUYmKJcP4RV2WFqKdC Instructor Hsiu-Hsuan Wang(王秀軒) Find more at https://anthony-wss.github.io/ Chapters 00:00 存取投影片…

• Funding for COVID-19 Projects

  3/15/2020, 4:31:23 PM — sam+altman

  I’m trying to fund startups/projects helping with COVID-19, because it’s basically the one thing I know how to do that can help.  I think we will soon have enough testing capacity, so now I’d like to start funding more startups working on: Producing a lot of ventilators or mask…

• 📄 Timothy Snyder On Tyranny Twenty Lessons From The Twentieth Century Tim Duggan Books (2017)

  6/8/2025, 10:06:11 PM — Books(local file)

  .EPUB file (4.4 MB) from Books - Modified: 2025-06-08

• Lenacapavir: The miracle drug that could end AIDS

  6/11/2025, 11:00:00 AM — wip_podcast

  Lenacapavir is a new HIV drug that blocks infections with an efficacy rate of nearly 100%, and it could completely change the fight against HIV worldwide. Saloni and Jacob talk about the development and prospects for this new drug, as well as the history of HIV, the initial disco…

• What I've Learned About AI in the Past Two Months.

  12/31/2024, 6:06:57 AM — sheracaolity

  This blog post was originally written in Chinese and translated into English entirely using ChatGPT o1. 如果想讀中文版，請滑到底。 Video interview version (Part 1). Video interview version (Part 2). Video interview version (Extra). Forward I probably don’t need to say much about it—AI is cha…

• 📄 Erich Fromm The Art Of Loving Thorsons (1995)

  2/21/2022, 1:40:45 AM — Exploratory-pdfs(local file)

  .PDF file (6.9 MB) from Exploratory-pdfs - Modified: 2022-02-21

• An addendum to the German fiscal austerity debates

  6/17/2025, 6:58:39 AM — marginal_revolution

  “There is a significant risk that France will be passed by neighbouring countries like Germany and Poland, who are working hard to increase military spending quickly,” said Tenenbaum. That is from the FT .  I am far from convinced that Germany will use its fiscal freedom wisel…

• 📄 Multimodal Clinical Benchmark For Emergency Care (mc Bec) A Comprehensive Benchmark For Evaluating Foundation Models In Emergency Medicine

  11/11/2024, 3:13:11 PM — Research Papers(local file)

  .PDF file (0.7 MB) from Research Papers - Modified: 2024-11-11

• How do declining fertility and climate change interact?

  6/24/2025, 6:24:45 AM — marginal_revolution

  There are lots of assumptions behind these results, but still it is good to see someone working through some scenarios: A smaller human population would emit less carbon, other things equal, but how large is the effect? Here we test the widely-shared view that an important bene…

• Germany Italy fact of the day

  6/23/2025, 6:04:17 PM — marginal_revolution

  Germany and Italy hold the world’s second- and third-largest national gold reserves after the US, with reserves of 3,352 tonnes and 2,452 tonnes, respectively, according to World Gold Council data. Both rely heavily on the New York Federal Reserve in Manhattan as a custodian, eac…

• 建議

  2/27/2025, 10:28:28 PM — sheracaolity

  不確定是什麼原因，但我發現過去半年，主動聯繫我、向我尋求人生建議的大學生變多了。這其實也合理，畢竟我也 27 歲了（天哪），而向比自己年長 10 歲內的人尋求建議，本來就是很常見的事。 然而，正是因為這些人來找我尋求建議，我才意識到這件事對我來說有多難。難的原因主要有兩個： 首先，我知道很多人在尋求建議時，其實只是想獲得一種「自己的選擇並沒有錯」的安定感。然而，正因為年齡漸長、看著過去的朋友和同學往不同道路發展，我才愈來愈能體會到人生的諸多面向，也愈來愈不再覺得有所謂正確的人生道路。舉例來說，我很喜歡我現在的生活，但我也知道有些人如果過著跟我一樣的生活絕…

• 📄 Do Vision Foundation Models Enhance Domain Generalization In Medical Image Segmentation

  11/14/2024, 3:15:20 PM — Research Papers(local file)

  .PDF file (0.7 MB) from Research Papers - Modified: 2024-11-14

• 📄 Thresholding For Making Classifiers Cost Sensitive

  11/28/2024, 1:15:05 AM — Research Papers(local file)

  .PDF file (0.3 MB) from Research Papers - Modified: 2024-11-28

• The Curious Case of the bos_token

  6/17/2025, 7:30:03 PM — lesswrong

  Published on June 17, 2025 7:00 PM GMT LLMs process inputs as a sequence of tokens. Typically, a dummy token is prepended to the sequence, known as the bos_token (beginning of sequence token).  Input: "Good morning"  Token strings: [' <|bos_token|> ', 'Good', ' morning']  Tokens:…

• Helion Needs You

  7/13/2022, 3:47:32 PM — sam+altman

  Helion has been progressing even faster than I expected and is on pace in 2024 to 1) demonstrate Q > 1 fusion and 2) resolve all questions needed to design a mass-producible fusion generator. The goals of the company are quite ambitious—clean, continuous energy for 1 …

• 📄 (trivia On Books) Trivion Books Zero To One Notes On Startups, Or How To Build The Future By Peter Thiel Trivia On Books (2016)

  8/5/2023, 12:22:44 AM — Books(local file)

  .EPUB file (0.2 MB) from Books - Modified: 2023-08-05

• 📄 Randal E. Bryant, David R. O’hallaron Computer Systems A Programmer’s Perspective (2015, Pearson) Libgen.li

  6/25/2025, 12:16:02 PM — Books(local file)

  .EPUB file (7.1 MB) from Books - Modified: 2025-06-25

• 開源項目thrift評價

  8/27/2009, 4:12:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>我可能是國內最早(08年)在實際項目(海內飯否)中部署thrift的人(之一)。 根據我自己的體驗給出如下介紹和評價<br/> 0 本質上他是一個跨語言序列化和反序列化封裝，網絡和服務器封裝，簡單RPC封裝，提供的功能和以下關鍵詞近：ACE，XML，joson，json，sokcet庫，CORBA，通信協議<br/> 1 實用：互聯網公司非常適合使用，見過很多公司自定的通訊協議非常不靈活和容易出錯。<br/> 2 輕便：雖然目前沒有正式文檔，但是閱讀...

• Interstellar travel will probably doom the long-term future

  6/25/2025, 3:32:26 PM — lesswrong

  Published on June 25, 2025 3:32 PM GMT I think people are very wrong about space colonisation increasing long-term existential security. Once we expand to other star systems, we may begin a self-propagating expansion of human civilisation throughout the galaxy. However, there are…

• CS 194/294-280 (Advanced LLM Agents) - Lecture 3, Yu Su

  2/18/2025, 5:36:47 PM — yt_deep_dives

• YC Alumni Reunion

  10/29/2024, 11:43:07 PM — sheracaolity

  這個月初，我到了一趟舊金山，主要是為了參加 YC 每年一度的 Alumni Reunion。我對舊金山的最後印象還停留在 2020 年疫情剛爆發、城市逐漸變成空城的模樣。當時離開時心裡是抱有一點遺憾的，因為有太多想在這邊做的事情來不及做，沒想到下一次回來已是近五年之後。雖然很多我以前讀書時會去的店都倒了，但是這裡那股難以言喻的創業氛圍倒是沒什麼改變。 這次回到舊金山，我也見了一些以前大學時的朋友。有趣的是，在台大認識的朋友大多選擇了南灣的科技巨頭，而在 Minerva 認識的朋友則更多投身於舊金山的新創。我找了三位 Minerva 的朋友一起 co-wo…

• What I Wish Someone Had Told Me

  12/21/2023, 10:44:25 PM — sam+altman

  Optimism, obsession, self-belief, raw horsepower and personal connections are how things get started. Cohesive teams, the right combination of calmness and urgency, and unreasonable commitment are how things get finished. Long-term orientation is in short supply; try not to wor…

• 抵制積極支持綠壩軟件廠商的電腦 轉貼

  6/12/2009, 7:07:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>@冉雲飛 工信部要求7月1日起銷售的電腦應安裝上網過濾軟件， PC廠商們反映，目前缺乏實施細則。方正、海爾公司態度積極，表態支持；聯想、戴爾、惠普態度謹慎，表示還需進一步瞭解情況。//請朋友們抵制方正、海爾對於安裝綠壩過於積極迎合的態度。<br/> 我支持對親少年上網進行過濾不良信息，但是堅決反對，這麼破的流氓軟件以保護奪去納稅人的錢 。</p> </div>

• 📄 Parameter Efficient Fine Tuning For Medical Image Analysis The Missed Opportunity

  3/1/2025, 6:04:27 PM — AI papers(local file)

  .PDF file (0.9 MB) from AI papers - Modified: 2025-03-01

• 關於成長的三個關鍵問題

  1/30/2025, 7:33:31 PM — sheracaolity

  過去半年因緣際會下更深入地認識了三家有 Strong Product Market Fit 的 SaaS 公司，有一些跟 Growth 有關的雜感想要記錄下來。 第一家公司已經成立超過 10 年，有一條穩定的產品線、營收和 2B 客戶群，10 年來成長時快時慢，有些客戶覺得他們的產品很棒，有些則覺得用處不大。這家公司在第 10 年的時候使用了某個新出現的技術在不到一季的時間內做了一個新的產品來解決他們原本的產品在解決的問題，這個產品讓既有客戶和新客戶們感到非常驚艷，所以這家公司營收開始高速自然成長，估值也因此翻了快 7 倍（接近 $1B）並決定出售。 第…

• 【生成式AI時代下的機器學習(2025)】第一講：一堂課搞懂生成式人工智慧的技術突破與未來發展

  2/23/2025, 10:23:57 AM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1ZEVgcwNBCqaPQlrAbjJb8S6BQJkEZYdp/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 感謝 Xyfer 同學提供字幕 12:02 此處應為 Gemini Flash Thinking (投影片上少打了 Thinking 這個字) 34:30 除了 Autoregressive Generation 還有其他策略，詳見：https://youtu.be/Q…

• Stanford CS25: V1 I Transformer Circuits, Induction Heads, In-Context Learning

  7/17/2022, 9:00:33 PM — yt_ai_fundamentals

  "Neural network parameters can be thought of as compiled computer programs. Somehow, they encode sophisticated algorithms, capable of things no human knows how to write a computer program to do. Mechanistic interpretability seeks to reverse engineer neural networks into human und…

• 人生第一場官司出判決了

  12/24/2009, 7:50:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>是第一次訴訟，庭審是自辯的，答辯狀也自己寫。<br/> 結果很不錯，遠好於預期，比我願意接受的調解還好，至少算是小勝。不過其實如果重新打一遍我覺得有一定可能完勝的。感謝另外一位律師老鄉，給我書寫答辯狀的指導和其它建議。 也要感謝愛人，她在法庭實習過，做過書記員，寫過判決，瞭解辯護的原則和要點，這個法學碩士的知識終於發揮了點作用。<br/> 分享經驗<br/> 1 打官司是按證據說話不是按事實說話，不要設想法官主持公道，要理解法官只依照證據按法規執行的...

• The Origins of Wokeness

  1/1/2025, 12:00:00 AM — paul_graham

• Agentic Interpretability: A Strategy Against Gradual Disempowerment

  6/17/2025, 3:10:06 PM — lesswrong

  Published on June 17, 2025 2:52 PM GMT Authors: Been Kim, John Hewitt, Neel Nanda, Noah Fiedel, Oyvind Tafjord Full paper on arXiv   We propose a research direction called  agentic interpretability . The idea of agentic interpretability stems from the observation that AI systems …

• Stanford CS25: V2 I Introduction to Transformers w/ Andrej Karpathy

  5/19/2023, 6:00:25 PM — yt_ai_fundamentals

  January 10, 2023 Introduction to Transformers Andrej Karpathy: https://karpathy.ai/ Since their introduction in 2017, transformers have revolutionized Natural Language Processing (NLP). Now, transformers are finding applications all over Deep Learning, be it computer vision (CV)…

• One possible reason why the skill premium is declining

  6/25/2025, 6:09:03 PM — marginal_revolution

  This is especially true for those jobs that require the rudimentary use of technology. Until relatively recently, many people could get to grips with a computer only by attending a university. Now everyone has a smartphone, meaning non-graduates are adept with tech, too. The cons…

• Programming Bottom-Up

  6/18/2025, 2:28:05 AM — paul_graham

• CS 194/294-280 (Advanced LLM Agents) - Lecture 7, Caiming Xiong

  3/20/2025, 7:09:01 PM — yt_deep_dives

• Fred Smith, RIP

  6/22/2025, 9:08:57 PM — marginal_revolution

  He founded FedEx, a company that before the internet truly was a big deal.  The plan for the company was based on an undergraduate economics paper .  At age thirty Fred was in deep trouble .  And “ In the early days of FedEx, when the company was struggling financially, Smith…

• 寫點《萬曆十五年》的讀後感和近期自我反省的事情

  8/26/2009, 10:57:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>很久沒寫博客了，嚴肅的寫作其實是思考和提煉自己的一部分，我訓練不足。<br/>過去三個月看的書中有一本叫《萬曆十五年》，期間有一章專門寫海瑞。我們都知道海瑞享譽古今，但是書中卻認爲他對明社稷貢獻並不大，去查了許多資料類不少類似的評判。海瑞自身肯定有原因，我想其中有一條就是不會合作和隱忍，用當下的話來說就是太sharp了。我們暫且不論海瑞的主張是否都是正確或者進步，但是不懂得隱忍、繞彎、合作、平衡導致他是一個不做壞事，但也不會做事的人。<br/>我也一...

• 📄 Rick Rubin The Creative Act A Way Of Being (2023, Penguin Publishing Group) Libgen.li

  3/31/2025, 6:30:13 PM — Books(local file)

  .EPUB file (2.2 MB) from Books - Modified: 2025-03-31

• GPT-4o

  5/13/2024, 5:39:38 PM — sam+altman

  There are two things from our announcement today I wanted to highlight. First, a key part of our mission is to put very capable AI tools in the hands of people for free (or at a great price). I am very proud that we’ve made the best model in the world available for free in ChatGP…

• Making Sense of Consciousness Part 1: Perceptual Awareness

  6/25/2025, 9:10:03 PM — lesswrong

  Published on June 25, 2025 9:10 PM GMT Midjourney, “Gabor patches, blindsight, visual cortex, lateral geniculate nucleus, consciousness and the brain” I admit to having been averse, for a long time, to thinking about “consciousness.” The very topic felt “woo” somehow; I was eve…

• The Right Kind of Stubborn

  7/1/2024, 12:00:00 AM — paul_graham

• 📄 Adaptformer Adapting Vision Transformers For Scalable Visual Recognition

  2/28/2025, 4:56:44 PM — AI papers(local file)

  .PDF file (0.9 MB) from AI papers - Modified: 2025-02-28

• 《Your Brain at Work》

  7/31/2024, 11:20:13 AM — sheracaolity

  最近因為公司團隊規模變大了，再加上產品方面展開了幾個比較複雜的專案，導致我常常覺得時間不夠用。為了更有效地提高自己和團隊的生產力，我上個週末讀了《Your Brain at Work》 這本書。 這本書是我的一個創業者朋友推薦給我的，書中的想法其實沒有到很讓人意外，知識量也不是說特別高，如果給三年前的我讀可能會沒什麼感覺，但是現在這個時機點讀起來反而意外地感到很多地方很受用，因為它給了我一個不錯的框架去檢視自己當前的做事方式和領導方式有沒有哪些地方值得調整，而答案確實是有的（例如我不該每天起床第一件事情就是處理 Email、我在什麼情況下不應該給員工反饋…

• Project Covalence

  6/16/2020, 5:08:44 PM — sam+altman

  Almost every company and non-profit working on COVID-19 that I offered to help asked for support with clinical trials—for companies focusing on developing novel drugs, vaccines, and diagnostics, rapidly spinning up trials is one of their biggest bottlenecks.  Science remains the …

• 逼自己提高

  11/12/2009, 9:05:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>不用逼自己無時不刻注意主動提高才是NB的人。當然之所以大家普遍不NB是其實本質懶惰和逃避，說到底就是懶惰。<br/> 有一幫夥伴可以互相督努力提高是一件很愜意的事。</p> </div>

• 招聘高級產品經理

  1/2/2010, 5:37:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>職位要求：<br/> -2年以上互聯網行業經驗；<br/> -具有敏銳的市場意識、良好的運營思維和創新意識；<br/> -善於溝通和協調資源，較強的組織能力和責任心；<br/> -對互聯網及相關行業有獨立的見解<br/> -具備計算機、互聯網相關知識技能<br/> 聰明而不裝b，緊貼事實而又前瞻，邏輯分析能力強，敏銳細緻，愛學習，喜歡分享和溝通<br/> 待遇有競爭力，期權股份優厚</p> </div>

• 📄 Mome Mixture Of Multimodal Experts For Cancer Survival Prediction

  6/24/2025, 6:19:50 PM — Research Papers(local file)

  .PDF file (1.0 MB) from Research Papers - Modified: 2025-06-24

• 【生成式AI時代下的機器學習(2025)】第十二講：語言模型如何學會說話 — 概述語音語言模型發展歷程

  5/24/2025, 7:12:58 PM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1JjbWkIMOWKmeouWtrMi7ph5-2j8TPlUF/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 如果想了解更多語音語言技術相關的內容，可以參考以下資料： Towards Holistic Evaluation of Large Audio-Language Models: A Comprehensive Survey https://arxiv.org/abs/250…

• 【生成式AI時代下的機器學習(2025)】第八講：大型語言模型的推理過程不用太長、夠用就好

  5/2/2025, 4:47:00 PM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1OovnEYdAocl8HUFx2hMvPX8dsLrqPdmd/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 字幕由網友 Xyfer 熱心提供 銜接上週課程 第七講：DeepSeek-R1 這類大型語言模型是如何進行「深度思考」（Reasoning）的？ https://youtu.be/bJFtcwLSNxI?si=_6QxjMfcJh4KNLvx

• Founder Mode

  9/1/2024, 12:00:00 AM — paul_graham

• The Effects of Ranked Choice Voting on Substantive Representation

  6/24/2025, 6:42:44 PM — marginal_revolution

  Ranked choice voting (RCV) is an increasingly popular electoral institution that has been posited by reformers and media outlets to produce transformative effects on electoral outcomes and representation. However, there is little social scientific evidence available that evaluate…

• Coming soon: The Works in Progress Podcast

  6/7/2025, 12:42:24 PM — wip_podcast

  Coming soon: the Works in Progress Podcast. Featuring underrated ideas to improve the world – for bigger, more beautiful cities; clean energy that's too cheap to meter; truly pathbreaking scientific research; everyday progress in things like food and drink; and more. Plus: Hard D…

• What Languages Fix

  6/18/2025, 2:28:05 AM — paul_graham

• 📄 Diagnose Like A Pathologist Transformer Enabled Hierarchical Attention Guided Multiple Instance Learning For Whole Slide Image Classification

  6/25/2025, 6:49:23 PM — Books(local file)

  .PDF file (3.8 MB) from Books - Modified: 2025-06-25

• 📄 Digital Profiling Of Gene Expression From Histology Images With Linearized Attention

  2/12/2025, 9:43:01 AM — Research Papers(local file)

  .PDF file (3.7 MB) from Research Papers - Modified: 2025-02-12

• 📄 李录谈现代化 从人类文明史角度看当今中美关系走向

  9/2/2024, 2:56:24 PM — Exploratory-pdfs(local file)

  .PDF file (1.0 MB) from Exploratory-pdfs - Modified: 2024-09-02

• Is there a looming Cultural Omnicide?

  6/25/2025, 6:23:52 PM — lesswrong

  Published on June 25, 2025 6:18 PM GMT The reality, risks, & consequences of scaling consumer-entertainment technologies to a global population with little cognitive security. [mirrored from Substack ] Western, American culture will be in frontier language models’ training data.…

• 📄 Multimodal Whole Slide Foundation Model For Pathology

  12/2/2024, 11:35:08 AM — Research Papers(local file)

  .PDF file (16.3 MB) from Research Papers - Modified: 2024-12-02

• How constrained is the NYC mayor?

  6/25/2025, 5:00:23 PM — marginal_revolution

  I thought to ask o3, here is the opening of its answer : New York City has a “strong-mayor / council” system, but the City Charter, state law and an array of watchdog institutions deliberately fragment power. In practice the mayor can move fastest on implementation —issuing e…

• Double Podcast Drop on AI Safety

  6/25/2025, 8:11:49 PM — lesswrong

  Published on June 25, 2025 8:11 PM GMT TL;DR Two days ago, I released two podcast episode: "One Big Bad Bill" on muckrAIkers ➡️ Igor Krawczuk and I discuss the substantial bill that Trump is attempting to push through the legislature, and I get a little animated...  (IMO it's j…

• How To Invest In Startups

  1/13/2020, 4:41:14 PM — sam+altman

  There is a lot of advice about how to be a good startup founder.  But there isn’t very much about how to be a good startup investor. Before going any further, I should point out that this is a particularly hard time to invest in startups—it’s easier right now to be a capital-take…

• CS 194/294-280 (Advanced LLM Agents) - Lecture 6, Ruslan Salakhutdinov

  3/11/2025, 1:25:16 AM — yt_deep_dives

• How to Do Great Work

  7/1/2023, 12:00:00 AM — paul_graham

• 議價能力

  11/12/2009, 8:54:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>學會讓對方不好意思拒絕，讓對方不能啓齒是一個能力。<br/> 其實方法在《影響力》這本書裏也有寫，但是理解了能否敢用會用忍心用又是一個能力，也是一個人生態度。<br/> 我議價能力中等偏下。</p> </div>

• AI Safety at the Frontier: Paper Highlights, May '25

  6/17/2025, 5:16:39 PM — lesswrong

  Published on June 17, 2025 5:16 PM GMT tl;dr Paper of the month: Models can detect when they're being evaluated with high accuracy, and potentially undermine safety assessments by behaving differently during testing versus deployment. Research highlights: LitmusValues measures ho…

• 📄 Nick Huber The Sweaty Startup How To Get Rich Doing Boring Things Harper Business (2025)

  6/8/2025, 10:07:34 PM — Books(local file)

  .EPUB file (3.7 MB) from Books - Modified: 2025-06-08

• Wednesday assorted links

  6/25/2025, 4:08:26 PM — marginal_revolution

  1. John Cochrane on central bank independence . 2. Possible progress on an AI-driven dengue vaccine? 3. Male and female literary genres . 4. “ Zohran Mamdani doing extremely well with college-educated voters, while the working-class overwhelmingly rejects him, is further…

• When To Do What You Love

  9/1/2024, 12:00:00 AM — paul_graham

• 📄 David Epstein Range How Generalists Triumph In A Specialized World (28 May 2019, Penguin Publishing Group) Libgen.li

  6/3/2025, 10:21:16 PM — Books(local file)

  .EPUB file (1.1 MB) from Books - Modified: 2025-06-03

• Writes and Write-Nots

  10/1/2024, 12:00:00 AM — paul_graham

• Stanford CS25: V2 I Emergent Abilities and Scaling in LLMs

  5/21/2023, 6:00:20 PM — yt_ai_fundamentals

  January 24, 2023 Emergent Abilities and Scaling in LLMs Jason Wei In this speaker series, we examine the details of how transformers work, and dive deep into the different kinds of transformers and how they're applied in different fields. We do this by inviting people at the for…

• o3 Turns Pro

  6/17/2025, 1:50:04 PM — lesswrong

  Published on June 17, 2025 1:50 PM GMT You can now have o3 throw vastly more compute at a given problem. That’s o3-pro. Should you have o3 throw vastly more compute at a given problem, if you are paying the $200/month subscription price for ChatGPT Pro? Should you pay the $200,…

• 【生成式AI時代下的機器學習(2025)】第五講：大型語言模型訓練方法「預訓練–對齊」(Pretrain-Alignment) 的強大與極限

  3/29/2025, 3:53:19 AM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1XeiZP-00kckqCxkgvi5QO47zyY1SAHqa/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 不熟悉大型語言模型訓練過程的同學可以先看以下影片： Pretrain: https://youtu.be/cCpErV7To2o?si=V1AmHnYmCAog5YiT SFT: https://youtu.be/Q9cNkUPXUB8?si=QuHhtTtipAs7VeH…

• 📄 [the Poppy War 1] R.f. Kuang The Poppy War (2018, Harper Voyager) Libgen.li

  6/19/2025, 11:48:10 PM — Books(local file)

  .EPUB file (1.6 MB) from Books - Modified: 2025-06-19

• Tuesday assorted links

  6/17/2025, 4:06:53 PM — marginal_revolution

  1. New Mennonite colony for Angola? (NYT) 2. Therapy loops for your AI? 3. Karl Smith Op-Ed return. 4. Rohit sentences to ponder, LLMs seem relatively safe . 5. New Musa al-Gharbi book is now under contract . 6. Conversations with Coleman podcast is now at The Fr…

• 軟和硬

  11/11/2009, 9:13:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>你不爭取別人不會因此感謝你，你爭取了別人也不會就不和你合作。<br/> 有的時候我顯得有點tough，有的時候又據說很是太軟弱不爭取，現在想來應該兩者都不是，應該是沒有在正確的時候選擇合適的態度。<br/> 今天聽到朋友討論一個糾紛，有人說到，算了我們還是別曝光了這是下下策。我想諸如此類事情：既然認爲自己是對的，你不曝光更多時候讓人覺得你是站不住理，或者有內情，同事“對手”也不會感謝你“爲他考慮”，相反你很堅決積極的態度反而能使大衆相信支持你讓“對手...

• 【生成式AI時代下的機器學習(2025)】第七講：DeepSeek-R1 這類大型語言模型是如何進行「深度思考」（Reasoning）的？

  4/25/2025, 9:00:07 PM — yt_ml_course

  投影片：https://docs.google.com/presentation/d/1lQwn837sXA9_sN0YFuPWScwFETbNbe6e/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 這堂課介紹了四種讓語言模型能夠進行深度思考的方法： (1) 更強的思維鏈（Chain-of-Thought，CoT） (2) 為模型提供推理工作流程 (3) 教導模型推理過程（模仿學習，Imitation Learning） (4) 以結果為導向學習推理（強化學習，Reinfo…

• 📄 Performance Analysis Of Cost Sensitive Learning Methods With Application To Imbalanced Medical Data

  11/28/2024, 1:44:58 AM — Research Papers(local file)

  .PDF file (1.6 MB) from Research Papers - Modified: 2024-11-28

• 📄 Diagnose Like A Pathologist Transformer Enabled Hierarchical Attention Guided Multiple Instance Learning For Whole Slide Image Classification

  6/24/2025, 6:19:44 PM — Research Papers(local file)

  .PDF file (3.8 MB) from Research Papers - Modified: 2025-06-24

• 【生成式AI時代下的機器學習(2025)】第九講：你這麽認這個評分系統幹什麽啊？談談有關大型語言模型評估的幾件事

  5/3/2025, 4:55:28 PM — yt_ml_course

  《生成式人工智慧導論 2024》也有介紹大型語言模型的評量，還沒看過的同學可以先去看那部影片 https://youtu.be/Hk8Z0uhmWg4?si=SbjzejHrwODYKRYT 字幕由網友 Xyfer 熱心提供 投影片連結：https://docs.google.com/presentation/d/1FKDNO2BWYl80PjkgpmCnr-7cbmSJ3DH8/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 18:30 上課時的解釋可能不夠精確，這裡再…

• 發郵件是競爭力，關於郵件推廣經驗總結

  5/24/2009, 10:44:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>上週和一個朋友聊天，他說去年幸虧做了定向的郵件推廣，否則業務就支撐不下去了。<br/>這兒不去討論郵件推廣的合法與道德，該與不該，怎樣纔算道德。僅僅從實踐的角度，從企業看到的效果出發來記錄一些總結。</p><p>效果： 郵件推廣效果，曾經比一個月化了10萬的百度競價排名還有效果很多</p><p>重要性：從重要性來講和SEO相當，主動的找到感興趣用戶，並且更詳細的內容介紹</p><p>使用者： 適合電子商務網站的產品推薦郵件，社區SNS的邀請郵件，普通...

• PG and Jessica

  9/25/2020, 2:45:50 PM — sam+altman

  A lot of people want to replicate YC in some other industry or some other place or with some other strategy. In general, people seem to assume that: 1) although there was some degree of mystery or luck about how YC got going, it can’t be that hard, and 2) if you can get it off th…

• DALL•E 2

  4/6/2022, 6:15:13 PM — sam+altman

  Today we did a research launch of DALL•E 2, a new AI tool that can create and edit images from natural language instructions.  Most importantly, we hope people love the tool and find it useful. For me, it’s the most delightful thing to play with we’ve created so far. I find it to…

• 最近半年讀過正在讀的書

  4/30/2009, 6:40:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>最近半年讀過正在讀的書</p><p>Seven Habits of Effective People<br/>Positioning<br/>活法<br/>我們爲什麼生病</p> </div>

• 【生成式AI時代下的機器學習(2025)】第二講：一堂課搞懂 AI Agent 的原理 (AI如何透過經驗調整行為、使用工具和做計劃)

  3/8/2025, 5:42:02 PM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1kTxukwlmx2Sc9H7aGPTiNiPdk4zN_NoH/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 5:45 此處應為 AlphaZero 而不是 AlphaGo Zero 6:10 此處並不是說 RL 不重要，畢竟 LLM 的訓練過程中也有使用 RL。此處想要表達的意思是，LLM 已經取得了某種程度的通用能力，不需要再針對不同的任務進行 RL 訓練。

• Expertise and determinism

  3/29/2024, 8:22:38 PM — sheracaolity

  When I encounter something I find difficult to understand or hard to do, I often find it intriguing and dedicate energy to figuring out how to understand and do it. However, once I've actually accomplished it and gained a certain level of expertise, my interest in 'how to do it' …

• 株洲事件 飯否實時搜索圖片

  5/18/2009, 5:08:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>女性的地位提高的真快<br/> 作者： onebird<br/> 發佈時間： 2009-04-30 19:12<br/> 分類： 生活</p><p>在微軟的team同事中已經有4個人已經或準備爲了老婆的博士學業到國外讀書或者工作。<br/> 完全跟着愛人的學習計劃。嗯，新女性，新時代。</p> </div>

• 知識陳舊

  9/4/2009, 7:01:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>最近訂購服務器，從朋友那瞭解了一些服務器的新技術，發現大學學的計算機體系結構中的知識不少已經old了。<br/> 可能也很難有書籍對新的體系結構作歸納總結的嗎，還是隻能看企業的文檔和博客文章。<br/> http://en.wikipedia.org/wiki/Intel_Nehalem_%28microarchitecture%29<br/> Nehalem</p> </div>

• How to Get New Ideas

  1/1/2023, 12:00:00 AM — paul_graham

• Three Observations

  2/9/2025, 9:05:32 PM — sam+altman

  Our mission is to ensure that AGI (Artificial General Intelligence) benefits all of humanity.  Systems that start to point to AGI* are coming into view, and so we think it’s important to understand the moment we are in. AGI is a weakly defined term, but generally speaking we mean…

• The Risk of Vertical AI

  3/11/2025, 3:16:46 PM — sheracaolity

  前陣子 YC 不斷在 拍影片 鼓勵創業者在選擇題目時多考慮做 Vertical AI，並分享了許多 Vertical AI 公司營收高速成長、迅速成為獨角獸的成功案例。然而，我對 Vertical AI 是否是一個好的創業題目，一直抱持很大的懷疑，以下是我的一些思考。 據我所知，相較於大廠的 General AI（例如：ChatGPT、Claude），當前市場中的 Vertical AI 有三個能力優勢： 資料操作 ：Vertical AI 能比 General AI 更有效地讀寫該 Vertical 的資料，例如 Cursor AI 能直接讀寫 Cod…

• The Practical Value of Flawed Models: A Response to titotal’s AI 2027 Critique

  6/25/2025, 10:15:27 PM — lesswrong

  Published on June 25, 2025 10:15 PM GMT Crossposted from my Substack .  @titotal recently posted an in-depth critique of AI 2027. I'm a fan of his work, and this post was, as expected, phenomenal*. Much of the critique targets the unjustified weirdness of the superexponential…

• Alfred Brendel, RIP

  6/17/2025, 5:59:01 PM — marginal_revolution

  His Haydn piano sonatas were his best work and also they were pathbreaking in helping us understand that composer.  I also very much appreciated his short Beethoven pieces on the old Vox box set, starting with the Bagatelles, and as a live performer his Schubert was memorable… …

• Stanford CS25: V2 I Language and Human Alignment

  5/20/2023, 6:00:04 PM — yt_ai_fundamentals

  January 17, 2023 Language and Human Alignment Jan Leike In this speaker series, we examine the details of how transformers work, and dive deep into the different kinds of transformers and how they're applied in different fields. We do this by inviting people at the forefront of …

• 轉一篇豆瓣的文章

  6/4/2009, 8:39:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p><a href="http://www.douban.com/group/topic/6792408/">http://www.douban.com/group/topic/6792408/</a></p><p>這篇日誌。送給我的飯否。</p><pre><code> 　　親愛的。你要快點好起來。 　　我們都想回來。 　　我們永遠愛你。 ...

• CS 194/294-280 (Advanced LLM Agents) - Lecture 5, Charles Sutton

  3/10/2025, 9:49:43 PM — yt_deep_dives

• Watch R1 "think" with animated chains of thought

  6/17/2025, 10:38:06 AM — lesswrong

  Published on June 17, 2025 10:38 AM GMT [This spent a couple days on top of HackerNews in February; see here for discussion. Best used as an loose visualization of LLM reasoning. Note that the distances used for the bar and line charts are actual cosine sim, not tSNE artifacts.…

• Helion

  11/5/2021, 1:39:16 PM — sam+altman

  I’m delighted to be investing more in Helion . Helion is by far the most promising approach to fusion I’ve seen. David and Chris are two of the most impressive founders and builders (in the sense of building fusion machines, in addition to building companies!) I have ever met, a…

• How To Be Successful

  1/24/2019, 9:01:07 PM — sam+altman

  I’ve observed thousands of founders and thought a lot about what it takes to make a huge amount of money or to create something important. Usually, people start off wanting the former and end up wanting the latter. Here are 13 thoughts about how to achieve such outlier success.…

• The Gentle Singularity

  6/10/2025, 9:12:47 PM — sam+altman

  We are past the event horizon; the takeoff has started. Humanity is close to building digital superintelligence, and at least so far it’s much less weird than it seems like it should be. Robots are not yet walking the streets, nor are most of us talking to AI all day. People stil…

• 創業的三目標

  12/20/2009, 7:01:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>創業過程於我有三個目標：事業有成，修煉自己，團隊磨練 。<br/>事業有成：希望能做出業界有巨大影響力的產品、公司、有豐厚的回報。<br/>修煉自己：希望在體能健康、知識技能、情感情商、精神境界 都有鍛鍊。<br/>磨練團隊：能夠達成有能共同進步，互相指正，默契配合，有所專長的團隊。<br/>很多時候不少人只專注在第一項，這樣很容易隨着事情的起伏而失去追求。</p> </div>

• Serving LLM on Huawei CloudMatrix

  6/17/2025, 5:59:50 AM — lesswrong

  Published on June 17, 2025 5:59 AM GMT AI 2027 Compute Forecast basically completely ignores China for Compute Production section and I don't think it can be justified. This paper from Huawei is a timely reminder.  Discuss

• Idea Generation

  5/28/2020, 7:12:40 PM — sam+altman

  The most common question prospective startup founders ask is how to get ideas for startups. The second most common question is if you have any ideas for their startup. But giving founders an idea almost always doesn’t work. Having ideas is among the most important qualities for a…

• A Methodologist's Apology

  6/25/2025, 4:52:22 PM — lesswrong

  Published on June 25, 2025 4:52 PM GMT I shall ask, then, why is it really worth while to make a serious study of mathematics? What is the proper justification of a mathematician’s life? And my answers will be, for the most part, such as are expected from a mathematician: I think…

• 📄 Generative Ai To Augment The Fairness Of Foundation Models In Cancer Pathology Diagnosis.

  6/19/2025, 7:52:02 PM — Research Papers(local file)

  .PDF file (0.1 MB) from Research Papers - Modified: 2025-06-19

• CodeReivew

  9/27/2009, 8:20:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>因爲最近工作原因今晚和在不同公司的老同事一起review了一個晚上的一個老系統的代碼，感想如下。<br/> 1 往回看，總是發現以前的工作做的很多問題，不夠好，蠻醜的。<br/> 2 原來的系統搞的挺複雜的，對老東家的技術團隊表示擔心。<br/> 3 要不斷回顧過去，不斷總結，多交流<br/> 4 不同公司的人一起codereview其實有蠻多交流，但是因爲商業原因難有機會，交流設計的機會都不多。</p> </div>

• 【生成式AI時代下的機器學習(2025)】第三講：AI 的腦科學 — 語言模型內部運作機制剖析 (解析單一神經元到整群神經元的運作機制、如何讓語言模型說出自己的內心世界)

  3/14/2025, 6:46:02 PM — yt_ml_course

  投影片連結：https://docs.google.com/presentation/d/1dmR4HtycPwofpOEgmdjqMdlTsTtfYcFR/edit?usp=sharing&ouid=115046073158939078465&rtpof=true&sd=true 12:10 神經元的『啟動』定義並非只有一種，例如並不一定要輸出值大於 0 才算啟動，也可以將輸出值大於平均值視為啟動。 59:30 此處應為「減去」功能向量，「加上」為口誤 1:10:05 此處應為「y double prime」，「w double prime」為口…

• Practice what you preach

  6/17/2025, 6:13:29 PM — marginal_revolution

  From the University of Barcelona: Master’s Degree in Political Ecology Degrowth and Environmental Justice By the way, the web site uses cookies. The post Practice what you preach appeared first on Marginal REVOLUTION .               Comments on the bright side, the …

• The Strength of Being Misunderstood

  12/1/2020, 6:56:39 PM — sam+altman

  A founder recently asked me how to stop caring what other people think. I didn’t have an answer, and after reflecting on it more, I think it's the wrong question. Almost everyone cares what someone thinks (though caring what everyone thinks is definitely a mistake), and it's prob…

• The Deadly Cost of Ideological Medicine

  6/17/2025, 11:17:38 AM — marginal_revolution

  Excellent Megan McArdle column in the Washington Post tracing how we have swung from one form of insanity on vaccine policy to another with barely a pause in between: In more than 20 years of covering policy, I have witnessed some crazy stuff. But one episode towers above the…

• Tuesday assorted links

  6/24/2025, 3:57:57 PM — marginal_revolution

  1. Arc Institute releases the first draft of a virtual cell model . 2. The continuing resurgence of Fischer Black in current macro . 3. Robert Nelsen on why the diabetes breakthrough is important . 4.  Law Review Puts Out Full Issue Of Articles Written With AI . 5. Ch…

• Heptabase August 2024 AMA

  8/26/2024, 4:59:36 PM — sheracaolity

  I hosted an AMA in Heptabase 's Discord community last week. This post includes my responses to the top upvoted questions, which cover the following topics: Heptabase's position in the PKM market and its attitude toward AI. Heptabase's attitude toward interoperability. Heptabas…

• 📄 A Survey Of Pathology Foundation Model Progress And Future Directions 2025 May

  6/18/2025, 11:54:01 PM — Research Papers(local file)

  .PDF file (0.7 MB) from Research Papers - Modified: 2025-06-18

• Stanford CS25: V2 I Biomedical Transformers

  5/25/2023, 6:00:20 PM — yt_ai_fundamentals

  February 21, 2023 Biomedical Transformers Vivek Natarajan In this speaker series, we examine the details of how transformers work, and dive deep into the different kinds of transformers and how they're applied in different fields. We do this by inviting people at the forefront o…

• 從打官司談起

  9/3/2009, 4:13:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>昨天打完人生第一個官司，還是自辯的。雖然是小官司，但是也是成熟路上的一步 。<br/> 每個人走出大學校園常見會因租房、找工作、辭職發生社會關係衝突，剛畢業的學生都很擔心，一來覺得無助，二來經濟條件有限承擔不起風險。其實社會關係上的因利衝突是不可避免的，租賃、僱用、合夥、買賣、合作關係都容易起糾紛。糾紛解決途徑因情況而異，法律是衝突解決常見手段途徑之一，具備一些知識準備，心理準備，人脈準備很必要。<br/> 租房糾紛可能是最常見的，剛出校園就會租房，...

• 📄 The Straussian Moment

  2/23/2025, 12:06:03 AM — Exploratory-pdfs(local file)

  .PDF file (0.5 MB) from Exploratory-pdfs - Modified: 2025-02-23

• Fictional Thinking and Real Thinking

  6/17/2025, 7:13:05 PM — lesswrong

  Published on June 17, 2025 7:13 PM GMT Background Concept: What World Does The Referent Live In? When we see some symbols representing something, it’s useful to ask “what world does the referent [1]  live in?”. Some examples: When I’m reading a Harry Potter book, the referents of…

• 酷訊被收購了

  10/30/2009, 10:55:00 PM — onebird

  <div class="post-content" itemprop="articleBody"> <p>半個月前猜到了，今天正式公佈，，很多原同事來找我問期權還是否管用（問我怎麼管用？），不管期權是否管用，從頭開始參與創業的公司賣了數千萬美金價格至少還是很沾光的。<br/> 其實我想說的是大家既然選擇創業公司工作就應該學習一下創業公司的相關知識，瞭解 一下liquidation preference， vesting ，和期權行權。<br/> http://www.feld.com/wp/archives/2005/01/term-sheet-liqu...

• Have Appliances Declined in Durability?

  6/23/2025, 11:17:22 AM — marginal_revolution

  Many Americans believe that their appliances have become less durable and reliable over recent decades. Rachel Wharton at Wirecutter has an excellent piece pushing back. Her conclusions mirror what I found when looking at clothing quality : yes, there has been a modest decline…

• 雙螢幕

  6/29/2024, 10:49:33 AM — sheracaolity

  2024/06/03 最近在設計師同事蠱惑之下，終於在創業第三年買了一個 27 吋的外接螢幕，把 Heptabase 接上去後直接看到新世界，覺得人生突然間變彩色了。 結果今天我在跟工程師同事 1-1 時說起這個新螢幕如何提高了我工作效率時，同事不以為然地跟我說他有三個螢幕：「啊這樣寫 Code 比較方便啊」 原來是小弟孤陋寡聞、不思進取、安於現狀。以後不敢說自己是做生產力工具的創辦人了，今天回去會好好自我檢討 Orz 2024/06/23 一直以來我都是在床上做事效率特高的人，每次東西想不透時，我要不就是去洗澡、要不就是拿著筆電到床上工作，大腦的轉速會…

• doubanclaim649a7916a3faff18

  7/28/2009, 6:44:00 AM — onebird

  <div class="post-content" itemprop="articleBody"> <p>---<br/> 穿支持google的tshirt出門 穿滾蛋gfw的thirt出門 讓你封網 封了我就把想說的話寫在衣服上<br/> 作者： onebird<br/> 發佈時間： 2009-06-25 06:22<br/> 分類： 生活</p><p>用內部實習員工來僞造採訪的CCAV<br/> 人工僞造用戶查詢來構造suggestion陷害谷歌<br/> 對關係到人們生活工作學習 的世界第一大網站google說封就封<br/> 再這樣下去  還有...

• 📄 Paul Millerd The Pathless Path (2022, Paul Millerd) Libgen.li

  3/31/2025, 6:30:38 PM — Books(local file)

  .EPUB file (2.9 MB) from Books - Modified: 2025-03-31

• What to Do

  3/1/2025, 12:00:00 AM — paul_graham

• 📄 Minerva Hcs Intro

  6/14/2025, 11:24:35 PM — Exploratory-pdfs(local file)

  .PDF file (0.3 MB) from Exploratory-pdfs - Modified: 2025-06-14

• 📄 Deep Learning In Cancer Genomics And Histopatholog

  2/7/2025, 3:27:40 PM — Research Papers(local file)

  .PDF file (1.8 MB) from Research Papers - Modified: 2025-02-07

• 📄 Entrepreneur Press Start Your Own Business, Fifth Edition The Only Start Up Book You'll Ever Need Entrepreneur Press (2010)

  6/8/2025, 9:41:03 PM — Books(local file)

  .EPUB file (2.4 MB) from Books - Modified: 2025-06-08

• 📄 Lora Low Rank Adaptation Of Large Language Models

  11/27/2024, 9:29:25 PM — AI papers(local file)

  .PDF file (1.5 MB) from AI papers - Modified: 2024-11-27

• 📄 Molecular Driven Foundation Model For Oncologic Pathology

  2/11/2025, 2:22:39 PM — Research Papers(local file)

  .PDF file (3.6 MB) from Research Papers - Modified: 2025-02-11

• Tech for Thinking

  6/25/2025, 9:30:48 PM — lesswrong

  Published on June 25, 2025 9:30 PM GMT My post on Neutrality was kind of abstract, and I think it’s time to do a follow-up that’s more concrete. Specifically, the place where I think there’s something to be done is in information technology, and the ways it shapes the perceived…

• 📄 The Design Of Everyday Things

  8/14/2024, 6:56:20 PM — Exploratory-pdfs(local file)

  .PDF file (2.3 MB) from Exploratory-pdfs - Modified: 2024-08-14

• 📄 Good Strategy Bad Strategy

  8/21/2024, 8:56:19 PM — Exploratory-pdfs(local file)

  .PDF file (2.1 MB) from Exploratory-pdfs - Modified: 2024-08-21

• 📄 Virchow2 Scaling Self Supervised Mixed Magnification Models In Pathology

  1/2/2025, 1:59:35 AM — Research Papers(local file)

  .PDF file (1.1 MB) from Research Papers - Modified: 2025-01-02

• CS 194/294-280 (Advanced LLM Agents) - Lecture 10, Sean Welleck

  4/17/2025, 11:42:06 PM — yt_deep_dives

• 📄 Fairtune Optimizing Parameter Efficient Fine Tuning For Fairness In Medical Image Analysis

  2/21/2025, 12:43:24 AM — Research Papers(local file)

  .PDF file (0.9 MB) from Research Papers - Modified: 2025-02-21

• My Paris delta

  6/24/2025, 4:48:01 AM — marginal_revolution

  I have not been here since 2019, so here are the trends I am noticing: 1. Vastly more shops are open on Sundays than before. 2. Central Paris continues to evolve into a nearly bilingual city.  It is not quite Amsterdam or Stockholm, but getting there.  And the Parisians do no…

• The Reddits

  3/1/2024, 12:00:00 AM — paul_graham

• 創業筆記(六): 聚焦與招聘

  5/31/2025, 10:15:07 AM — sheracaolity

  最近突然發現，距離上一次發表創業筆記系列的文章竟然已經過了三年。三年前公司還處於從 0 到 1 的階段，因此當時寫的主題大多圍繞在打造產品、成長營收，以及一些早期新創在成立過程中可能會經歷的事情（找尋共同創辦人、設立公司、申請 YC、種子輪籌資）。 如今三年已過，公司早已脫離了從 0 到 1 的階段，秉信著「經驗在當下分享更有價值」的原則，我決定更新一下這個系列，分享我在創業從 1 到 10 的階段所學到的事情。今天分享的主題會著重在「聚焦」和「招聘」這二個面向，希望對其他處在相同階段的創業者有所幫助。 聚焦 我認為，一個新創 CEO 在從 0 到 1（…

• CS 194/294-280 (Advanced LLM Agents) - Lecture 8, Thomas Hubert

  3/31/2025, 7:19:44 PM — yt_deep_dives

• 📄 Sft Memorizes, Rl Generalizes A Comparative Study Of Foundation Model Post Training

  6/13/2025, 3:34:37 PM — AI papers(local file)

  .PDF file (4.6 MB) from AI papers - Modified: 2025-06-13

• CS 194/294-280 (Advanced LLM Agents) - Lecture 9, Kaiyu Yang

  4/13/2025, 1:56:14 AM — yt_deep_dives

• Reinforcement Learning Progress

  6/25/2018, 4:10:37 PM — sam+altman

  Today, OpenAI released a new result .  We used PPO (Proximal Policy Optimization), a general reinforcement learning algorithm invented by OpenAI, to train a team of 5 agents to play Dota and beat semi-pros. This is the game that to me feels closest to the real world and comple…

• Stanford CS25: V2 I Common Sense Reasoning

  5/24/2023, 6:00:20 PM — yt_ai_fundamentals

  February 14, 2023 Common Sense Reasoning Yejin Choi In this speaker series, we examine the details of how transformers work, and dive deep into the different kinds of transformers and how they're applied in different fields. We do this by inviting people at the forefront of tran…

• Why Arc Isn't Especially Object-Oriented

  6/18/2025, 2:28:05 AM — paul_graham