Jirat C
20 followers · 23 following · 1813 views
on the atlas — 167
- Nothing in Biology Makes Sense Except in the Light of Evolution1 savers
- Theistic evolution1 savers
- Vitalism1 savers
- Emergentism1 savers
- Horses3 savers
- Why Reactionaries Are Taking Over the World - The Atlantic1 savers
- You can't reach the brain through the ears3 savers
- Peoples, Armies and Governments of WWII (notes) | writing1 savers
- Relax for the same result | Derek Sivers6 savers
- Why I Love Mormonism - The New York Times1 savers
- Dispatches from the possibly last days of human relevance (Shtetl-Optimized)3 savers
- Searching for God in Silicon Valley - by Avital Balwit5 savers
- Chip design from the bottom up – Reiner Pope - YouTube6 savers
- Building something no one else can measure3 savers
- Model capabilities and four minute mile1 savers
- AutoGo: a Tutorial3 savers
- PersonalSit.es | Yes we got hot and fresh sites3 savers
- Every site needs a Links Page / Why linking matters | Melon's Thoughts - https://thoughts.melonking.net/thoughts/every-site-needs-a-links-page-why-linking-matters2 savers
- The first thing I did last year was run | Henry From Online2 savers
- aman.ai • the art of artificial intelligence3 savers
- THE METIS LIST4 savers
- Interconnects AI | Nathan Lambert | Substack3 savers
- Yi Tay2 savers
- Freshman Year — Founders School1 savers
- Gemini Flash Pretraining2 savers
- best of SF (tarlon's version)2 savers
- [2310.17813] A Spectral Condition for Feature Learning2 savers
- Tempering hardcoreness - by Ajeya Cotra - Good Bones5 savers
- Neurosciences Seminar: X. William Yang, MD, PhD - Beyond the inherited CAG: Decoding a neuronal-selective pathogenic cascade in Huntington's disease | Wu Tsai Neurosciences Institute1 savers
- Should you marry her? - by Ajeya Cotra - Good Bones3 savers
- All non-drone militaries are obsolete4 savers
- alternbits/awesome-cuda-books: A curated list of best cuda programming books ·1 savers
- Free Alpha By Design | near.blog5 savers
- How to Buy Cheap Claude Tokens in China - by Zilan Qian8 savers
- ProgramBench1 savers
- Deriving Muon17 savers
- What Ethiopian runners taught me about reading scientific literature2 savers
- The advice I would give on a mentorship call10 savers
- Memory access is O(N^[1/3])3 savers
- In My Misanthropy Era — LessWrong2 savers
- Real-Time Action Chunking with Large Models2 savers
- Radical fun - by Ava - bookbear express4 savers
- Looking for Alice - by Henrik Karlsson - Escaping Flatland90 savers
- Even if you beat me - The Dublin Review58 savers
- Nat Friedman57 savers
- sparkly people and how to find them - by Anson Yu53 savers
- An Opinionated Guide to ML Research52 savers
- You don’t need to work on hard problems51 savers
- Details That Make Interfaces Feel Better43 savers
- advice - nabeelqu43 savers
- I should have loved biology36 savers
- How To Scale Your Model34 savers
- Learn In Public30 savers
- How to write a cold email27 savers
- How Superhuman Built an Engine to Find Product/Market Fit | First Round Review26 savers
- Rest in motion24 savers
- How to win a best paper award (or, an opinionated take on how to do important research)24 savers
- The New World - Colossus24 savers
- no exit - by Ava - bookbear express23 savers
- Post 38: On Slack - Having room to be excited — Neel Nanda23 savers
- Interaction Models: A Scalable Approach to Human-AI Collaboration - Thinking Machines Lab23 savers
- The Gentle Singularity - Sam Altman21 savers
- Willingness to look stupid20 savers
- Advice to Young People, The Lies I Tell Myself - jxnl.co19 savers
- Galaxy brain resistance18 savers
- To be a Technologist is to be Human - Letters to a Young Technologist18 savers
- Why Don't We Just Kill the Kid In the Omelas Hole by Isabel J. Kim : Clarkesworld Magazine – Science Fiction & Fantasy17 savers
- Top Performers are Pathologically Ambitious - by Matt Beard17 savers
- The Second Half – Shunyu Yao – 姚顺雨17 savers
- Transformer Circuits Thread16 savers
- A Website To End All Websites | Henry From Online16 savers
- Are you serious? - by visakan veerasamy16 savers
- Greg - Sam Altman16 savers
- Stanford’s War on Social Life15 savers
- Shtetl-Optimized » Blog Archive » The First Law of Complexodynamics15 savers
- Universal Love, Said The Cactus Person15 savers
- Advice to Young People, The Lies I Tell Myself - jxnl.co14 savers
- math team - by benedict - bene dictio14 savers
- Gossip is Good13 savers
- bell labs13 savers
- The Annotated Transformer13 savers
- jasmine sun's visit to china12 savers
- Nonfiction Writing Advice | Slate Star Codex12 savers
- Career advice | What's new12 savers
- LLM Daydreaming · Gwern.net12 savers
- The Longevity FAQ12 savers
- Examples of barbell strategies11 savers
- Digital hygiene | karpathy11 savers
- Modal11 savers
- AGI Trades11 savers
- Gradual Disempowerment11 savers
- fall 23 syllabussy | arielle lok10 savers
- Structured Procrastination10 savers
- 21 observations from people watching - by shani10 savers
- Nobody Is Perfect, Everything Is Commensurable | Slate Star Codex10 savers
- Matt Huang | Bitcoin for the Open-Minded Skeptic10 savers
- i have found a way out - by Sophie Kim10 savers
- Current AIs seem pretty misaligned to me — LessWrong9 savers
- Pandoc - About pandoc9 savers
- What I Wish I Knew in College9 savers
highlights — 145
How can we contrive to be at once astonished at the world and yet at home in it?
Searching for God in Silicon Valley - by Avital BalwitThe madman is not the man who has lost his reason. The madman is the man who has lost everything except his reason.
Searching for God in Silicon Valley - by Avital BalwitI became so engrossed with regaling myself, preening over my own glorious victory, that I almost did not react in time for the road abruptly ending.
The first thing I did last year was run | Henry From OnlineBut relating to something from a place of inadequacy introduces all sorts of blind spots. You’ll be biased to not seriously entertain counterarguments that threaten your new source of self-worth (e.g., dismissing critics who claim your work is actually causing harm in X way). You also end up judging others harshly based on whether they ‘get it’, since your judging yourself harshly on this.
Tempering hardcoreness - by Ajeya Cotra - Good BonesAnd people are fundamentally tolerant of one another’s personal choices, and resist engaging in purity death-spirals
Tempering hardcoreness - by Ajeya Cotra - Good Boneswhere they elicit proteotoxicity leading to profound chromatin defects. The latter include erosion of epigenomic signatures of MSN identity
Neurosciences Seminar: X. William Yang, MD, PhD - Beyond the inherited CAG: Decoding a neuronal-selective pathogenic cascade in Huntington's disease | Wu Tsai Neurosciences Institutewhere I actually couldn’t live with myself is the one where I took this thing that was a perfect match for everything I’ve always naturally loved and cared about, and been fortunate to have some aptitude for, and just didn’t give it my all
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusMore seriously, I think this shift from survival mode to creative mode is the right one. It’s not going to be in the next two or four years, but it’s coming. At some point you really will only be capped by what you can imagine and the ideas you have for how you want to spend your life, which will be a very different life. To your question, I think a lot of it will be pure self-expression. People will get to live the lives they want to live. The bottleneck to turning our dreams into reality will dramatically narrow. And the way we live now will seem crazy.
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusIn some sense, solving AI really does mean solving cognition.
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusI kind of like to say that retrieval is one of the things we really do best, which is actually kind of to your point about memory.
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusHowever, I think of it less as a direct application and more as a proof of concept of what’s possible. In order to do some of these really hard problems, you need to develop a lot of fundamental logical reasoning, a lot of creativity, and the ability to string together very, very long chains of reasoning and make sure they all line up. Knowing that models can do this with sufficient reinforcement learning and the right training data does suggest, pretty clearly, that a lot of these practical tasks can be done.
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusIt’s in some ways a wonder that humans can do math problems at all.
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusWe’ve always lived in Minecraft survival mode, and now we’re going to be living in creative mode
The Wu Tapes: Q&A with Cognition's Scott Wu - ColossusI know it’s trite, but this is exactly how I feel when I see (for example) a paper that uses category theory to prove NL≠NP. We start in Boston, we end up in Beijing, and at no point is anything resembling an ocean ever crossed.
Shtetl-Optimized » Blog Archive » Ten Signs a Claimed Mathematical Breakthrough is WrongHow do we make a public AI debate substantive? How do we ensure policymakers can make informed decisions about AI, so that we can sensibly regulate some of the negative consequences of AI without stifling the positive uses?
How AI Will Reshape Computer Systems by 2035: A Jeffersonian Dinner in San Francisco about Our 10,000x Future - CRA-IIt is our job to ensure that we develop and encourage the societally- beneficial aspects of AI, and we discussed education and healthcare as two domains with considerable early positive benefits and enormous further potential
How AI Will Reshape Computer Systems by 2035: A Jeffersonian Dinner in San Francisco about Our 10,000x Future - CRA-ISecond, there is an existence proof that we can do much better since humans learning from birth to early adulthood are able to use 1,000x less input data than today’s largest ML models, suggesting that much more data-efficient learning algorithms are still possible
How AI Will Reshape Computer Systems by 2035: A Jeffersonian Dinner in San Francisco about Our 10,000x Future - CRA-IWe conjecture that in five years 10,000x more AI inference will be done worldwide, with these gains hypothetically coming from multiplicative progress of 50x in AI algorithms, 50x in system/hardware optimization/specialization, and 4x from further data center growth.
How AI Will Reshape Computer Systems by 2035: A Jeffersonian Dinner in San Francisco about Our 10,000x Future - CRA-I“Every story I create, creates me. I write to create myself.” —Octavia E. Butler
FSPO: Few-Shot Optimization of Synthetic Preferences Personalizes to Real UsersFiltering questions by LLM labeled difficulty or LLM response length yields better results than filters typical to pre-training data curation that use embeddings or fastText
OpenThoughts3 - A new SOTA Reasoning Data Recipe | Open ThoughtsMany alternative names are exonyms in the practice of making new infectious diseases seem foreign.[36][37][38] This pattern was observed even before the 1889–1890 pandemic, also known as the 'Russian flu', when the Russians already called epidemic influenza the 'Chinese catarrh', the Germans called it the 'Russian pest', and the Italians called it the 'German disease'.[39][40] These epithets were re-used in the 1918 pandemic, along with new ones.[41]
Spanish fluThey are more pregnant with meaning than any stanza of poetry: each word provides a logical toehold I can mount after extracting everything implicit in the preceding ones
Understand - a novelette by Ted Chiangthey could resist if they wished, but so few of them do.
Understand - a novelette by Ted ChiangBlinding, joyous, fearful symmetry surrounds me.
Understand - a novelette by Ted ChiangMy new language is taking shape. It is gestalt-oriented, rendering it beautifully suited for thought, but impractical for writing or speech.
Understand - a novelette by Ted ChiangSharpening. RL compresses variance and stabilizes known strategies: pass@1↑ without expanding the support at large K. Discovery. RL crosses a learnability gap: pass@K=0 → sustained solves after training, with a distinct phase transition in the learning curve. Which regime you get depends on reward topology, exploration duration, data mixture, and task hardness.
Center for Responsible, Decentralized Intelligence at BerkeleyIn grug language: Smart man sees big problem. Often state of nature is many small things. Smart man make one big thing out of many small things to throw at big problem. But then evil man take big thing from smart man and make more problem. Or big thing grow legs and beat smart man without making problem go away. This is bad. Maybe better to throw small things at big problem and not make big thing, even if solve problem less. Or before make big thing have plan for how to not have big thing do evil.
Do not conquer what you cannot defend — LessWrong“Civilization is, in its earliest phases, played. It does not come from play like a baby detaching itself from the womb: it arises in and as play, and never leaves it.”
Grimes and the Game Economy – Drew's Inkhaven Web Logfortifying such an effort would be extremely difficult; the technology itself is destabilising and risks a loss of human control; and geopolitical adversaries would pre-emptively sabotage projects
Dan Hendrycks warns America against launching a Manhattan Project for AIAs Geoffrey Hinton, an ai pioneer, warned with characteristic understatement, “There is not a good track record of less intelligent things controlling things of greater intelligence.”
Dan Hendrycks warns America against launching a Manhattan Project for AIranging from the Hopfield model and Boltzmann machine (2024 physics Nobel prize), and the understanding of optimization dynamics and geometry in high dimensional disordered systems (2021 physics Nobel prize),
Home | Physics Of LearningTime spent isn’t an end goal and long hours are often a vanity metric - things that make you look good in front of others but don’t help you achieve the goal.
It doesn’t matter how many hours you workBird that I am, I don't know where I need to go so much as that I need to. Not from this city, but from this work and this state of mind. I'd like to think that we're not entirely the same, though. The birds fly far away because they think the seasons are going to make their homes and habits unlivable; their anxiety to migrate is almost a form of generational trauma. But I could live like this forever. I'd like to think what drives me is almost a generational optimism. I have a sense that I can winter anywhere if need be, that the trade winds fluttering for me thousands of miles away are pecul…
Zugunruhe - Aidan SmithMindy called me a hero. But I was simply a man in the right place at the right time. Dr. Hamilton is also a hero because he designed the Hopeful. Mindy is also a hero because she kept me awake. My mother is also a hero because she was willing to give me up so that I could survive. My father is also a hero because he showed me the right thing to do.
Mono no aware - Lightspeed MagazineThe easy question, I think, is how the change should find us. In 1948 CS Lewis said “Let that bomb, when it comes, find us doing sensible and human things – praying, working, teaching, reading, listening to music, [etc…]” This ethos is sensible for most people (who are unable to contribute to solving these problems), but it is an unacceptable answer for those working on AI today. If you wake up every morning and try to improve the trajectory of the transformative AI systems that could remake the world, you have an obligation. What percentage of humanity gets to influence these decisions, even …
Top Performers are Pathologically Ambitious - by Matt BeardThis is the person I got to know: curious, lively, sharp as a razor, and fun.
Jim Simons - Emmanuel CandèsJim: This does sound crazy. Should I believe you? Emmanuel: This is a mathematical fact. Jim: Do you have a proof? Emmanuel: I think so. Jim: I would like to see it.
Jim Simons - Emmanuel Candès“And finally you drive to the top of the highest peak and you find a sage, and you ask him what series of buttons on the dashboard you have to press to get out of the car. And he tells you that it’s not about pressing buttons on the dashboard and you just need to GET OUT OF THE CAR. And you say okay, fine, but what series of buttons will lead to you getting out of the car, and he says no, really, you need to stop thinking about dashboard buttons and GET OUT OF THE CAR. And you tell him maybe if the sage helps you change your oil or rotates your tires or something then it will improve your driv…
Universal Love, Said The Cactus PersonThe efficient market hypothesis is a lie At best it is a very lossy heuristic The best things in life occur where EMH is wrong In many cases it's more accurate to model the world as 500 people than 8 billion "Most people are other peo
Nat FriedmanIt's important to do things fast You learn more per unit time because you make contact with reality more frequently Going fast makes you focus on what's important; there's no time for bullshit "Slow is fake" A week is 2% of the year Time is the denominator
Nat FriedmanThird, labour income can rise under partial automation because automation scales the value of remaining bottleneck tasks.
O-Ring Automation - Marginal REVOLUTION“Chess is a game of skirmishes,” I say. “The perspective of Go is bigger. It encompasses entire battles.” “There are no heroes in Go,” Bobby says, stubbornly.
Mono no aware - Lightspeed MagazineUnderstand that we are not defined by our individual loneliness, but by the web of relationships in which we’re enmeshed. A person must rise above his selfish needs so that all of us can live in harmony. The individual is small and powerless, but bound tightly together, as a whole, the Japanese nation is invincible.”
Mono no aware - Lightspeed MagazineTrillions of dollars is going to tap out every source of venture capital, every investor who is willing to put in money now for an uncertain but potentially huge future return. To get trillions of dollars you will need boring debt investors, people who need their money back, people who are willing to put in money now for 9% a year and repayment in five years. Very different psychology.
Musk Mars Merger, Maybe - Bloomberg[Art history is] stereotypically a field for prep school graduates, especially women, with plenty of family wealth to fall back on. In fact, a New York Times analysis of Census data shows that art history majors are wildly overrepresented among those in the top 1 percent of incomes. Perhaps the causality runs from art history to high incomes, but I doubt it.
Musk Mars Merger, Maybe - BloombergAstrology-based investing
Musk Mars Merger, Maybe - BloombergFinance clubs I was a classics major in college, focused on reading Greek and Latin poetry. Somewhat later, I became an investment banker. That was a thing that people used to do. You treated college as a sort of luxury consumption good, you read some poetry, you had some fun, you lazed about on the quad, you graduated, you were like “ugh job I guess,” and you got a job. Some kinds of jobs (doctor) were only available to people who spent long years studying the relevant subject matter. Most kinds of jobs, though, were open to bright hard-working young people with good general educations and a …
First Brands Did Some Round Trips - BloombergFunds get more fragile with scale. So building competitive advantage becomes existential if you want to build an institution that endures. The problem is, that isn’t how fund managers are encouraged to spend their time or their focus. Most funds are run by an alpha decision maker who oversees all investments. They spend most of their time thinking about the next marginal deal, and not much time thinking about their moats. Compensation structures reward investment returns, split among small teams. Firms, on the other hand, are run by entrepreneurs. And entrepreneurs think constantly about compe…
Firm > Fund - by David Haber - a16zEven the people who love me don’t exist solely to love me; they have their own hangups and priorities and self-concept, and they must do what they can live with. This was a bitter lesson that I learned this year, but there was also an extreme sweetness to it, because in learning it I also realized: Other people are real, but so am I, and I am whole as I am, flawed and selfish and imperfect as I am. When I realized this, it was like I had been granted full permission to exist for the very first time.
nothing will change your life more than saying what you actually want to sayMy one goal of 2026, which has been my goal every year since 2020, is to be zen at all costs. As a formerly neurotic person I’ve come far on this front. I now have a reputation in my friend group as someone who is comfortable with sharing whatever deranged thought crosses her mind—I will wake up at 8:30 AM and send a loved one a three paragraph text about how I’m feeling paranoid. I can’t tell you how weird that feels to admit, as someone who used to be extremely focused on how everything was received. The reason I was obsessed with how someone received my message was because I cared so much a…
nothing will change your life more than saying what you actually want to say