Eric Yoon

on the atlas · 47

• Brain dump on the diversity of AI risk | Jascha’s blog - https://sohl-dickstein.github.io/2023/09/10/diversity-ai-risk.html3 savers
• Opinion | What One Russian Satellite Tells Us About the Future of Nuclear Warfare - The New York Times - https://www.nytimes.com/interactive/2024/12/05/opinion/nuclear-weapons-space.html1 savers
• A Gentle Introduction to the Rectified Linear Unit (ReLU) - MachineLearningMastery.com2 savers
• ross perot policies - Google Search - https://www.google.com/search?q=ross+perot+policies&oq=ross+perot+policies&sourceid=chrome&ie=UTF-81 savers
• Can Solar Costs Keep Shrinking? - by Tomas Pueyo2 savers
• The Friendship That Made Google Huge17 savers
• Alan Eustace - Wikipedia - https://en.wikipedia.org/wiki/Alan_Eustace1 savers
• Yale Law Journal - Amazon’s Antitrust Paradox4 savers
• jlevy/the-art-of-command-line: Master the command line, in one page - https://github.com/jlevy/the-art-of-command-line?tab=readme-ov-file1 savers
• Price Adapter - https://www.priceadapter.com/2 savers
• Commonplace Book - https://arvid.xyz/commonplace-book/5 savers
• The Technium: You Are Not Late - https://kk.org/thetechnium/you-are-not-late/10 savers
• The Architecture of Open Source Applications2 savers
• The Technium: Class 1 / Class 2 Problems - https://kk.org/thetechnium/class-1-class-2-problems/4 savers
• Bringing Generative AI to Healthcare | Sequoia Capital - https://www.sequoiacap.com/article/generative-ai-for-healthcare-perspective/?itm_medium=related-content&itm_source=sequoiacap.com1 savers
• Introduction | Uncurled - https://un.curl.dev/intro1 savers
• Copying is the way design works || Matthew Ström: designer & developer2 savers
• spruce.world - https://spruce.world/blog/grand-unified-theory-of-sf.html2 savers
• spruce.world - https://spruce.world/blog/why-not-to-drop-out.html2 savers
• Wong Kar-wai - Google Search - https://www.google.com/search?sca_esv=9aab8662baee94c4&sxsrf=ADLYWIKyJANSS6GkTGF9DcK_EaWk72lbjg:1721786513813&q=Wong+Kar-wai&si=ACC90nzx_D3_zUKRnpAjmO0UBLNxnt7EyN4YYdru6U3bxLI-LwxhoK4LUSU2GaOObjWO0Fen2pkhZxwcNyOR65Da_Uuqfc1AGqI_GyH7Zqg0OVwtZvNQzZe9uI3SDVOeWB2-eOuz_UB9EvSraYfZt5jXU1Ft23fN63oBiXsyYRwiHJgk7Gt1gKWLiKCqdLXKpGKoFX5pO-vr&sa=X&ved=2ahUKEwjrqqPMyr6HAxW9FVkFHY9SDfQQmxMoAHoECBkQAg&biw=2277&bih=1412&dpr=1.51 savers
• Leslie Cheung - Wikipedia - https://en.wikipedia.org/wiki/Leslie_Cheung1 savers
• New Arrivals - For Sale at 1stDibs - https://www.1stdibs.com/new-arrivals/1 savers
• Alameda & FTX Archive - https://caroline.milkyeggs.com/1 savers
• An Intuitive Guide To Exponential Functions & e – BetterExplained - https://betterexplained.com/articles/an-intuitive-guide-to-exponential-functions-e/#:~:text=e%20is%20the%20base%20rate,growing%20just%20a%20little%20bit.1 savers
• A Guide to Quantization in LLMs | Symbl.ai - https://symbl.ai/developers/blog/a-guide-to-quantization-in-llms/1 savers
• Topological Sort - https://www.clear.rice.edu/comp314/lec/week6/Topological.htm1 savers
• From prototype to production: Vector databases in generative AI applications - Stack Overflow2 savers
• Claude’s Character \ Anthropic - https://www.anthropic.com/news/claude-character1 savers
• Too much efficiency makes everything worse: overfitting and the strong version of Goodhart’s law | Jascha’s blog9 savers
• Why GPS Is Under Attack - The New York Times - https://www.nytimes.com/interactive/2024/07/02/world/gps-threats.html1 savers
• Curius / Bookmarks for the extremely curious190 savers
• Even if you beat me - The Dublin Review58 savers
• How to Do Great Work51 savers
• Taste is a guide for what is worthwhile49 savers
• The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.37 savers
• these are the days - beauty and terror29 savers
• Do I Need to Go to University? -- colah's blog26 savers
• How to Learn Better in the Digital Age25 savers
• Bret Victor, beast of burden24 savers
• James Mulholland21 savers
• Tyler Cowen is the best curator of talent in the world21 savers
• Unhinged: Curiosity as Ambition — NPC #78820 savers
• The Zoomer Question - American Affairs Journal10 savers
• Big companies v. startups7 savers
• Deep Reinforcement Learning: Pong from Pixels5 savers
• J. Robert Oppenheimer3 savers
• Numbers Every Programmer Should Know By Year3 savers

highlights · 103

• Share knowledge or capabilities directly, without going through a slow and costly teaching process. If an AI model gains a skill, that skill can be shared by copying the model's parameters. Humans are unable to gain new skills by copying patterns of neural connectivity from each other.

  Brain dump on the diversity of AI risk | Jascha’s blog - https://sohl-dickstein.github.io/2023/09/10/diversity-ai-risk.html

• to bake the “understanding” of other relevant words into the one we’re currently processing.

  The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.

• hidden state allows an RNN to incorporate its representation of previous words/vectors it has processed with the current one it’s processing.

  The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.

• The encoder’s inputs first flow through a self-attention layer – a layer that helps the encoder look at other words in the input sentence as it encodes a specific word. We’ll look closer at self-attention later in the post.

  The Illustrated Transformer – Jay Alammar – Visualizing machine learning one concept at a time.

• And that’s the biggest mistake I think young people make: They’re afraid to look dumb, so they follow safe paths that cap their downside, not realizing that they also cap their upside.

  Take Asymmetric Bets - by Erik Torenberg - Ideas & Musings

• However, the bigger risk is not that you fail, it's that you don't get enough actual shots on goal to do something amazing.

  Reconsidering Career Optionality - Erik Torenberg

• The most successful people I know are primarily internally driven; they do what they do to impress themselves and because they feel compelled to make something happen in the world. After you’ve made enough money to buy whatever you want and gotten enough social status that it stops being fun to get more, this is the only force I know of that will continue to drive you to higher levels of performance.

  Sam Altman

• But it’s hard to trust when you are exposed to one of the most frightening of revelations: that hardly anyone ever bothers to check.

  Age of Invention: Does History have a Replication Crisis?

• They don't tell you, but I will: when it comes to figuring out what to work on, you're on your own.

  How to Do Great Work

• The three most powerful motives are curiosity, delight, and the desire to do something impressive.

  How to Do Great Work

• learn enough to get to the frontier, notice gaps, explore promising ones.

  How to Do Great Work

• Boldly chase outlier ideas, even if other people aren't interested in them — in fact, especially if they aren't.

  How to Do Great Work

• The way to figure out what to work on is by working. If you're not sure what to work on, guess.

  How to Do Great Work

• In fact, the correlation between superlinear returns and inequality is so strong that it yields another heuristic for finding work of this type: look for fields where a few big winners outperform everyone else.

  Superlinear Returns

• Which is another reason to start taking risks early.

  Superlinear Returns

• Choose work you have a natural aptitude for and a deep interest in. Develop a habit of working on your own projects; it doesn't matter what they are so long as you find them excitingly ambitious. Work as hard as you can without burning out, and this will eventually bring you to one of the frontiers of knowledge. These look smooth from a distance, but up close they're full of gaps. Notice and explore such gaps, and if you're lucky one will expand into a whole new field. Take as much risk as you can afford; if you're not failing occasionally you're probably being too conservative. Seek out the b…

  Superlinear Returns

• m I working on what I most want to work on? When you're young it's ok if the answer is sometimes no

  How to Do Great Work

• Try to finish what you start, though, even if it turns out to be more work than you expected.

  How to Do Great Work

• if you stick to what you find genuinely interesting, you'll be proof against all of them.

  How to Do Great Work

• There are a lot of forces that will lead you astray when you're trying to figure out what to work on. Pretentiousness, fashion, fear, money, politics, other people's wishes, eminent frauds.

  How to Do Great Work

• One sign that you're suited for some kind of work is when you like even the parts that other people find tedious or frightening.

  How to Do Great Work

• A popular student, he was vice president of his sophomore class, and president of his junior class. During his senior year, he managed the football team

  Vannevar Bush

• Bush was a well-known policymaker and public intellectual during World War II, when he was in effect the first presidential science advisor.

  Vannevar Bush

• The memex and Bush's 1945 essay "As We May Think" influenced generations of computer scientists, who drew inspiration from his vision of the future.

  Vannevar Bush

• dean of the MIT School of Engineering

  Vannevar Bush

• founded the company that became the Raytheon Company

  Vannevar Bush

• the creation of the National Science Foundation.[2]

  Vannevar Bush

• was an American engineer, inventor and science administrator, who during World War II headed the U.S. Office of Scientific Research and Development (OSRD),

  Vannevar Bush

• including important developments in radar

  Vannevar Bush

• In the early 20th century, the number of patients residing in mental hospitals increased significantly[n 2] while little in the way of effective medical treatment was available

  Lobotomy

• Most people fell into an intermediate group, left with some improvement of their symptoms but also with emotional and intellectual deficits to which they made a better or worse adjustment.

  Lobotomy

• Portuguese neurologist António Egas Moniz, shared the Nobel Prize for Physiology or Medicine of 1949 for the "discovery of the therapeutic value of leucotomy in certain psychoses

  Lobotomy

• severing connections in the brain's prefrontal cortex.

  Lobotomy

• with the pizza stone

  Bagels

• n my experience, a KitchenAid produces fine results,

  Bagels

• By this point, the dough should be very smooth, satiny yet slightly tacky (it's fine if it sticks to the bowl of the stand mixer, or your cutting board), and stiff.

  Bagels

• mix at speed 2 in increments of 2 minutes for a total of 8 minutes

  Bagels

• Then add 655g flour, 8g diastatic malt, and 20g salt

  Bagels

• I also have a really intense mixer, a pizza stone, a kitchen scale, and bagel boards.

  Bagels

• I think that bytes are particularly interesting because a single byte can represent 256 different combinations. (Remember powers of 2? 2 to the power of 8 is 256.) And what about if you have two bytes? Two bytes means 16 bits (binary digits), which means that you can represent 65,536 different combinations (2 to the power of 16)!

  Bits, Bytes, Building With Binary

• A single digit in binary is known as a binary digit. But, you might know it as a bit.

  Bits, Bytes, Building With Binary

• DNA methyltransferases

  DNA Methylation and Its Basic Function | Neuropsychopharmacology - https://www.nature.com/articles/npp2012112

• inhibiting the binding of transcription factor(s) to DNA

  DNA Methylation and Its Basic Function | Neuropsychopharmacology - https://www.nature.com/articles/npp2012112

• recruiting proteins involved in gene repression

  DNA Methylation and Its Basic Function | Neuropsychopharmacology - https://www.nature.com/articles/npp2012112

• of the cytosine to form 5-methylcytosine

  DNA Methylation and Its Basic Function | Neuropsychopharmacology - https://www.nature.com/articles/npp2012112

• When you have set spell enabled within your conf file, misspelled words are automatically underlined for you. You can also enable or disable this by running :set spell and :set nospell.

  Learn Vim For the Last Time - https://danielmiessler.com/p/vim

• # Go to the beginning of the line and enter insert mode

  Learn Vim For the Last Time - https://danielmiessler.com/p/vim

• Let’s recall our language: Verb, Modifier, Noun

  Learn Vim For the Last Time - https://danielmiessler.com/p/vim

• gg: go to the top of the file G: go to the bottom of the file

  Learn Vim For the Last Time - https://danielmiessler.com/p/vim

• }: move forward one paragraph

  Learn Vim For the Last Time - https://danielmiessler.com/p/vim