
George T 🦇🔊
754 posts

George T 🦇🔊
@georgety
Fan of cryptography and UX. Front End Developer by day. Dreamer by night.



New paper & surprising result. LLMs transmit traits to other models via hidden signals in data. Datasets consisting only of 3-digit numbers can transmit a love for owls, or evil tendencies. 🧵

Yes, modern founders will be half influencer half founder I think This isn't completely new, look at Steve Jobs or Elon Musk





I read an HN thread where Nvidia employees reveal how Jensen Huang runs the company. Here's the tl;dr: 1. While Jensen doesn't do 1:1s, he has placed managers at different organizational levels who do 1:1s with their subordinates. For example, an org2 manager hosts regular 1:1s with their org3 colleagues. ---------- 2. Each org has a top 5 list (things people are working on) to which employees send weekly updates. Jensen samples from these mailing lists and reads them every morning. By reading the employees’ priorities, he understands if their actions align with the company’s goal. The best part? Everyone in the org can read them. One can also subscribe to other org's mailing lists. ---------- 3. Jensen conducts regular group meetings. The goal is to have a ground truth of what’s happening in the company and ensure everyone has that context. One employee shared, "I was only a couple of months into working at the company before I got to have a face-to-face discussion with him about a project I was working on. Many mid-level engineers (IC4-IC5) give him deep dives in these group meetings. It can be very stressful being under Jensen's microscope, but it dramatically reduces the "let's show pretty slides to the CEO to show him everything is good" BS." ---------- 4. Jensen changes course quickly when presented with opposing facts. He once asked his employees to quit working on a project they spent months on. Why? CES, a competitor, developed the same product that would have taken Nvidia another two years to launch theirs. He announced the decision in a meeting, saying, “Every morning when I wake up, I look in the mirror and check if the core assumptions on which I based my decisions are still valid. If not, I make the change immediately." ---------- 5. Nvidia doesn’t have 1-year or 5-year plans. But “what folks work on are much longer-term visions. Everyone has/owns their own ideas and plans for how an area, product/service, or project should grow into, and they communicate those. The company adapts well to changing circumstances and market needs, but it’s strategic and thoughtful - once it goes after something, it rarely backs off. And, usually, NVIDIA is going after things others actively avoid because the problems in those spaces are too difficult. The consequence of such problem sets are visions that, if not more, 10-20 year undertakings that everyone chips away at bit by bit until one fine day things start taking solid shape.” My personal favourite are #2 and #4. The ability to laser-focus and change course quickly in such a large org is a superpower.



























