Meanwhile, 98% of fa/tv/irgins can't perform basic mathematics.
Have you heard back from those Nigerian Princes yet?
>>223617554Which reminds me, I need a shit
>>223617686Tried camembert a week ago. By god it smells like actual unwashed ass.
>>223617554affleck wondamon lost
Ben should stop drinking
>>223617699Based cheese connoisseur
>>223617699He looks better
My boy's wicked smaht
>>223617554His explanation is so far off it's hilarious. "pretty shitty Python scripts" is definitely underselling how bad he is at it.t. PhD deep learning from Yoshua Bengio's lab.
>>223617554He's so full of shit.
>>223617554so that's why he's such a mediocre actor
>>223617764Most his explanation is okay though.
>>223617764>Terrence Howard's lablol. lmao even.
>>223617554>bragging about shit you learn in the first year as undergrad for any IT field.
>>223617816First years are doing classes on algorithms and mathematics. ML is a second or third year elective.
I genuinely don't know anything beyond addition, subtraction, multiplying and division. Was never taught anything in school. Parents never cared to bother either.
>>223617764He's pretty spot on, maybe except that CNNs did not evolve into transformers >PhD from Bengio's labLmao, let me guess, you got it from MILA, and you carefully avoided saying he was your supervisor. And as if you have any connection to him at all, MILA is over a thousand people.
>>223617554>I've worked with GPUs a lot too.no one says this Ben is making shit up
>actor larpinghehe
>>223617842You could reach the level of a STEM undergrad in just a few months of driven study.Schools teach mathematics so incredibly inefficiently I'm not sure it's worth it at all.They should just focus on important cultural knowledge desu. Having the historical awareness to understand your political situation is more important than knowing which part of a linear equation is the y-intercept.
>>223617868I actually always have to say "from bengio's lab" because nobody knows what "MILA" is nowadays.I got my PhD back in the days right before he sold out (in circa 2017, at which point he started only giving all his students combined a 1h a month session). In those days it was constant close communication and lab presentations and 1:1 contact so yes, I know him well.>He's pretty spot onMy fucking sides!
>>223617842Check out mathisfun and Khan academy for the basics, you'll learn much faster
didn't openai just solve a shit ton of math problems? /sci/ pepes tell me mathkeks are having a meltdown
>>223618004This. Once you have the basics, you can spend 2 hours and do 2 trimesters of college-level courses at a mastery level you wouldn't be able to get from the best unis (why? Because teaching is bottom-of-mind for these "people")
>>223618036No, all bullshit marketing shit as usual.
>>223618036You mean they stole other people’s existing work on the math problems, fed it to AI and then claimed AI “solved” it on its own when it came up with the answer after doing the necessary amounts of calculations.
>>223618087what's that mean? that the problems were already solved and the academics were like waiting on a time slot on a supercomputer or something to get the proofs?
>>223617554I wish I could defend pisslam and have a trooned out kid as well.Ben's the goat!
>>223618149It means the AI didn’t just sit down and one day decide to solve a mathematical problem. Instead it was fed works by mathematicians who had been working on the problem, likely without any permission, and then had the AI process all the work that had already been done and after enough time it produced a solution. It’s like building a house for a lengthy period of time but then some random asshole comes in and nails the last nail in place and then claims they built the entire thing all on their own.
>>223617554Tell me your secrets, American pigdogs
>>223618149Most of these problems were "solved" by doing what amounts to writing a 30 lines python script that bruteforces every option until something hits, and letting it run for an hour to find a hit. The reason they were "not solved" is nobody ever gave a shit to even look at the problem in the first place. "Solving" the problem also changes nothing, but that's 99% of math.Other cases were just stolen from mathematician's chat sessions with AI tools. The navier-stokes-related millenium prize for example. But they didn't even solve it in the end even by lifting the existing solution, they solved an easier version than asked by the prize (and the conclusion was... everything people were doing in the past 40 years should be done the way it's been done for 40 years... woow...).In other cases, they claimed to solve problem X when they were actually solving problem Y which is the irrelevant cousin of problem X that was so useless it was never posed as a "problem" in the first place. For example, if problem X is about finding a tighter upper bound for something, the "AI" will find that "the existing LOWER bound is not tight" (but won't find a new lower bound either).
>>223618228>>223618238that's lame, surely it's at least good for SOMETHING
>>223618532>that's lame, surely it's at least good for SOMETHINGEvery single time I look into where I could use LLMs, that is word for word my reaction.
