@CosmicMultivac@burkov I wholeheartedly agree with other people pointing out about how unnecessarily snarky this person is, but modern day gauss is super wild
Terence Tao for some reason referenced ChatGPT as the source of some content in his paper.
Expect scientists to cite spelling correctors: "By the way, the word 'elucidate' was suggested by Microsoft Word spelling corrector. I wasn't previously aware of 'elucidate'."
I first tried to read this book in 2018 and couldn't make it through because I thought it was too hard.
8 years later it's the only book I recommend every developer reads, and I had the chance to review the 2nd edition.
Join a study group, give it a read.
@Inquisition1776@JoJoFromJerz@RpsAgainstTrump You should look at all the evidence and make up your own mind. I suspect that you dismissed the evidence because you supported him. I question your character.
Trump: “To be a great nation, you must have religion, & you must have God. In churches across the nation on Sunday, the pews will be fuller, younger, and more faithful than they have at any time in many, many years. Religion is growing again in our country”
@WalshFreedom You’ve abandoned conservative principles and ideals many years ago—so it’s no surprise conservatives will react that way.
But the Dems? Also no surprise they would act that way, too—because they insist you toe the line.
Me: “I’m a proud Zionist, a huge supporter of Israel, and I do NOT believe Israel committed genocide, and I do NOT believe Israel is an apartheid state.”
The Left: “Fuck you Joe, I’m done following you.”
Me: “I oppose Trump’s stupid, illegal war against Iran, and I oppose Netanyahu.”
The MAGA Right: “Fuck you Joe, I’m done following you.”
New art project.
Train and inference GPT in 243 lines of pure, dependency-free Python. This is the *full* algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further.
gist.github.com/karpathy/8627f…
Our grad-level "Deep Learning" course (MIT's 6.7960) is now freely available online through OpenCourseWare: ocw.mit.edu/courses/6-7960…
Lecture videos, psets, and readings are all provided.
Had a lot of fun teaching this with @sarameghanbeery and @jxbz!
@pmddomingos There are fresh comp sci graduates coming out of good schools who are still looking for jobs. I don't know how wise your advice is. Can you elaborate?
Hugo Duminil-Copin, French mathematician and 2022 Field Medalist told me he never participated in math competition and was very bad at it.
Innovative mathematics requires creativity, intuition, intense concentration, and long reflections, sometimes spread over several years.
Good performance at a math olympiad merely tests fast problem solving abilities. AI can do that nowadays.
One of the big activities of a researcher, in mathematics and elsewhere, is not to answer questions but to ask the right questions.
GPT-5.3-Codex is here!
*Best coding performance (57% SWE-Bench Pro, 76% TerminalBench 2.0, 64% OSWorld).
*Mid-task steerability and live updates during tasks.
*Faster! Less than half the tokens of 5.2-Codex for same tasks, and >25% faster per token!
*Good computer use.
With my experience and everything I know, I could come to any mostly white-collar company, talk to people about what their job tasks are, and architect a way to replace some of those tasks with AI, saving between 20% and 35% of costs to the employer or increasing productivity by the same amount.
I could do that, but knowing how simple it is today, I feel zero motivation to do that.
@yacineMTB Opus is good on the surface, good to start a brand new project, but I have repeatedly called on Codex to fix Opus's errors and find solutions to problems (like jemalloc memory retention issues).