Tweet épinglé
"You just begin. You do the math. You solve one problem... and you solve the next one... and then the next. And If you solve enough problems, you get to come home."
English
Bill (e/λ)
3K posts
Bill (e/λ)
@billmdevs
Computer Scientist; AI researcher. Working on AGI. Finally mustered the courage to try to understand the universe and the nature of reality.
Canada Inscrit le Temmuz 2016
790 Abonnements338 Abonnés
@richa_lq @FoundersPodcast Digital version has been available for a while now. I think I saw it for the first time last November or December iirc
English
@FoundersPodcast how are y'all getting your hands on it before it came out?
English
I read these 400 pages in 2 days. Working on a Founders episode on it now:
English
Laminar flow is beautiful. One of the few instances where movement matches exactly to fps of observable reality
まだ面白い@madaomoshiroi
流体の教科書には必ず出てくるけど意外と出会わない層流。時が止まってるように見えるのすげえ
English
I understand what you are saying but still… That’s one of the reasons I am grateful to be at a stage where I can choose the people who I work with and like them. Being in contracting and consulting also helps as I don’t deal with those general “goals”
My company goals are geared towards my research mostly.
English
But in order to achieve that, you need to write some achievable goals in line with the company’s vision so that when you are measured against them at the end of the year, you remain employed. You don’t have to car about the goals personally, but you do need to take them seriously
𐌁𐌉Ᏽ 𐌕𐌉𐌌𐌉@OrevaZSN
I hate when a job asks you to set "goals" My only "goal" is to remain employed so I can continue to pay for my living expenses while finding fulfillment in the areas of my life unrelated to a day job.
English
How do you encode the reward function of "creativity"?
Is it something already represented in the brain at a systems level?
What does neuroscience say about the firings that give rise to creativity?
English
Bill (e/λ) retweeté
The divide is "science" and "engineering" both of which are methods required for research. Academia usually bottlenecks because nobody bothers doing engineering. Industry bottlenecks because nobody bothers doing science.
kache@yacineMTB
now that i'm doing actual research instead of just engineering i realize lecun was right and elon was wrong
English
Bill (e/λ) retweeté
@billmdevs @CJHandmer there is no minute wasted looking at the data
English
Bill (e/λ) retweeté
"Don't you want your grandchildren to experience the same hardships, struggle, scarcity, and short lives your forefathers did?"
Life for most of human history was nasty, brutish, and short. Changing that means mastery over nature, which means changing things.
I want the Solar System for my children and their children. For their descendants? The stars.
Luke Leisher@luke_leisher_
They’ll say some shit like “don’t you want your grandchildren to look up and see the same moon your forefathers did?” And the masses will happily vote for it.
English
@alz_zyd_ It’s actually realising that life is short that makes him push the game forward
English
When you are young, if you are talented and ambitious, you want to be Elon, you can't possibly imagine quitting the game early like Gates/Bezos. As you get older you realize just how short life is
LinaHua@Linahuaa
Elon getting viscously bullied as a kid, and rejected by his father has created a deep wound he describes as "rage demon" that drives him forward. Bill Gates and Jeff Bezos retired pretty early to play backgammon, party with Epstein, or court women with big celebrity parties. Elon's hunger, on the other hand, is endless, and so intense, that he takes bigger risks and fakes it till he makes it harder than anyone. In a different timeline, Elon might have gone bankrupt with Tesla at some point, with a damaged aura that would have stopped his ambitions. But in this time line, he went all-in like 4 times in a row, and hit the nuts on the river every time so far. However, among the top billionaires he's also the most likely to crash his net worth by 80%+
English
@yacinelearning I kinda knew it but I think I actually understood that principle when I saw @CJHandmer ways of arriving at a breakthrough on the Vesuvius challenge. (He did it by following the “crackles” on the scrolls)
English
whenever I talk to a researcher that worked both in cutting edge and practical applications they always like:
“yeah the research stuff is cool, but tbh most of the gains in industry is just going in and cleaning the data”
English
Bill (e/λ) retweeté
I think most people should consider becoming scientists.
While there is a real risk AI will eliminate jobs, conversely the need for models to understand the world around us through more and better data will increase exponentially.
AlphaFold, for example, would not have been possible if not for the work of thousands of grad students and postdocs painstakingly depositing well-curated datasets over decades of meticulous experimental work.
We still only understand a fraction of the physical world and beyond, and we’ll only be able to do so if we collect more data and put it into context.
Maybe this is a bit naive or utopian thinking, but humans are not designed to look through pages of receipts, read 100-page legal documents, or write reports and POCs no one will ever read. They are wired to explore, driven by curiosity, and have an innate desire to understand the world.
Many who are bored or feel a lack of purpose -- with the overhanging (and maybe exaggerated) doom and gloom of AI replacement -- should think about this.
English
@yacineMTB Did the same calculus but I was too much of a purist to go down the distributed systems route.
Loved playing with ML research in academic environments at that time.
English
Ten years ago, when I was still in university, I wrote a bot to scrape all indeed job postings and their expected salaries. I ran histograms on the total income I'd expect, and found correlations to skills. This is why I decided to focus on distributed systems instead of data / ML jobs. They simply made more money
Tips Excel@gudanglifehack
Claude Cowork can apply to 50 jobs in under 30 minutes. Here's how to set it up.
English
@george_ @YuvrajS9886 No worries. I thought I should ask. Any allocation of computer is good opportunity to move forward 🙂
English
@billmdevs @YuvrajS9886 Hey, thanks for the detailed response. At this time, I'm looking to support multiple projects, so I don't have the additional capacity to support your long-running bursty workload. Best of luck with your research
English
I have $7,000 in extra GPU credits lying around.
Building something interesting in AI/ML/research/open-source?
Please comment / dm me! I'd love to help. Limited to Waterloo ppl / ppl I've met irl.
Waterloo, Ontario 🇨🇦 English
This is the best example I have heard in a while about innovation through LLMs. I have some thoughts on how to get around this I use personally coming out soon
GIF
Simo Ryu@cloneofsimo
Its very possible that LLM trained on newtonian physics may never come up with relativity to explain cosmic scale gravity. In that case Einstein would have to intervene and solve it instead. But would he have had come up with it, assuming he offloaded all the physics problem solving to LLMs? I think this is serious problem. Undoubtedly many GOATs are only GOATs because they built all the intuition from problem solving themselves. Grothendieck famously reinvented measure theory from scratch when he was teenager. If people offload their RL envs they couldve used, to LLMs, we will never get the next Einstein
English
@george_ @YuvrajS9886 A100 80GB, bursty usage over 4-6 months (~3-4K GPU hours). Fine-tuning 70B biomedical models + multi-agent reasoning experiments at scale.
I have a detailed compute plan — want me to DM you the breakdown?
English
@billmdevs @YuvrajS9886 Do you know how much compute/memory you need - what gpu, and for how long?
English
Bill (e/λ) retweeté
If you’re an AI PhD student just starting out, don't be discouraged by the hype of "autoresearch" automating scientific discovery.
It won't. AutoML made the same big promises in 2017, and we all know how that turned out.
Ignore the noise. Master the fundamentals and learn to do research from first principles. Trends fade, but a solid foundation is how you will actually thrive.
English
@billmdevs @RMladek I keep forgetting and then not caring that this isn’t an anon account 🤣
English
