How much are you willing to bleed, to suffer for what you believe

Joined March 2022
Chris motionless optimising my life’s loss function one scream at a time. #MIW
5
215
Web crawlers /[•]\ #js x #css
920
7,852
807
63,398
2,261,514
Ahmad Farhan retweeted
ICLR submissions
11
55
8
989
68,796
Ahmad Farhan retweeted
not now honey, I'm doing frontier AI research
46
110
18
1,919
102,118
Ahmad Farhan retweeted
What is going on at ICLR? 60k? There's still an hour left 😭
8
11
8
209
19,928
Ahmad Farhan retweeted
i shed a tear reading this blog post by a deepseek kernel engineer
64
463
61
6,750
209,927
Ahmad Farhan retweeted
you'd be surprised how well this works
259
289
50
14,720
557,049
sf is insane man met a founder with 18 employees all 18 of them live in his apartment with him (2 bedroom place) they work 6-11-7 (6am to 11pm, 7 days a week) they have to get his permission to do anything besides work or sleep he also didn't want them wasting time dating people outside the company, so he matched them up internally (80% of employees are guys so a few of them now have to be gay) sf startup culture is wild man
439
207
124
9,433
934,354
Ahmad Farhan retweeted
why can the fly do a three point turn 😭
22
45
9
815
34,299
Ahmad Farhan retweeted
i trained the fly to solve rubiks cube is there anything this little guy can't do?
590
1,975
564
34,454
6,079,316
Tug of ducks
18
43
9
687
82,256
Ahmad Farhan retweeted
A good week to remind people that about a year ago, @tarungupta360 discovered that AI research agents (when asked to generate novel ideas) often plagiarized from existing work and skillfully reworded outputs to evade plagiarism detectors. arxiv.org/abs/2502.16487
2
14
129
6,161
🦆 🏀
41
61
6
1,093
135,442
Ahmad Farhan retweeted
Holy retard, guys, public keys are PUBLIC
41
18
4
844
89,899
Ahmad Farhan retweeted
We are stuck in a massive local optimum.
67
240
33
2,743
226,748
I’m a retired brain surgeon and I wanna tell you why you feel like life is passing by. I just need one minute to explain this to you so hang in there. Repetitive daily routines stop the brain from recording clear detailed memories. When every day looks the same, your brain does not track time the same as a clock does. It tracks change. If nothing new happens then days, months, and years start to blur together. This is know as Time compression. When life lacks variety, your brain compresses past days together, making it feel like time is sped up. To stop feeling like life is passing you by, add new experiences to your life. Break routine habits, like changing your daily home routine, taking a different route to work, or trying a new hobby. Schedule small trips or plan new adventures so your days do not blend together. Practice mindfulness by staying fully focused on what you are doing right now and not trying to race off to the next task. Keep a journal and write down happy memories or memorable parts of the day before you go to sleep. Take deep, calm breaths when you feel anxious or rushed to return your focus to the moment. You can and you will retrain your brain to not feel like life is passing you by by following these few simple measures.
646
9,312
680
62,834
7,709,087
Neil Movva (@neilmovva) started his career at Nvidia, working on GPUs and kernels, and has an unusually deep understanding of inference, from software to chips to power. We spend a lot of time on each of those layers, how they connect, and where the important tradeoffs are. What makes this conversation special is how detailed it is (like a 401-level class), yet Neil makes it remarkably clear and easy to follow. Today he runs Sail Research, a company building infrastructure for agents to make tokens as cheap as possible. We discuss: - Latency versus throughput - Why there are no bad chips, only bad pricing - The end of kernel engineering - Buying chips and power no one else wants - New chip architectures - Nvidia lore + his contrarian view of the company - Open source and the frontier labs I learned a ton. Enjoy! TIMESTAMPS 0:00 Intro 0:38 Building a “Token Factory” 4:21 The Future of Background Agents 13:09 Nvidia and the GPU Stack 23:27 Chips, Memory, and Transformers 36:14 The Future of AI Training Data 44:32 Chip Scarcity and Compute Arbitrage 52:44 Reinventing the AI Data Center 59:01 Power and the “Scavenger Strategy” 1:10:10 Open vs. Closed AI
146
1,192
159
10,410
4,639,872
Your future delivery dude
206
348
126
3,586
648,002
Ahmad Farhan retweeted
Replying to @TheHumanoidHub
Bro even had time to fix his hair after going down the steps.
6
6
1
165
5,044
Ahmad Farhan retweeted
Earth isn’t a cosmic lottery ticket — new models show our world was practically inevitable For decades, planetary scientists treated Earth’s existence like a rare jackpot: the right size, the right distance from the Sun, the right conditions for life — all somehow lining up by pure chance. New simulations presented at the Origins 2026 conference in Paris are rewriting that story. Planetary scientist Nader Haghighipour of the University of Hawaii and his team ran more than 1,000 high-fidelity simulations of the late stages of terrestrial planet formation. Unlike earlier models that started with carefully chosen “solar-system-like” setups, these runs began from completely random distributions of planetesimals and planetary embryos. No assumptions. Just physics. The result? A planet the size of Earth forming at roughly 1 astronomical unit (Earth’s actual distance from the Sun) emerges as a natural, recurring outcome. Venus appears about 28% of the time and often stays in or near the habitable zone. Mars shows up repeatedly as a smaller body near its real orbit. In short, our familiar planetary lineup isn’t a fluke — it’s what the laws of physics tend to produce in a protoplanetary disk with a realistic, non-uniform distribution of solid material. Haghighipour puts it bluntly: “There is no reason to believe that our Earth is a fluke.” Tiny differences in the starting conditions can still send systems down wildly different paths, but the pathway that leads to an Earth-like world is surprisingly robust. The same simulations also hint that Earth-sized (and slightly larger) planets in the habitable zones of Sun-like stars should be common across the galaxy. Detecting life on those worlds remains far beyond our current technology. Yet understanding exactly how Earth became habitable — without any special pleading — gives us a much clearer target. Life may not be rare. The conditions that allow it to arise may simply be what nature does when you let the physics run free.
9
28
4
159
21,964
Humanoid robots showed impressive speed on the track. But once obstacles were introduced, their weaknesses became much more visible.
21
78
14
467
26,338