jebarker

29 ms·

421.
by jebarker 2y ago
The most important part of FIRE is avoiding lifestyle inflation, otherwise you're just treading water. On the other hand I've also seen folks retire early and then return to big tech because they didn't have anything to retir
422.
by jebarker 2y ago
> on Twitter It's like rain on your wedding day
423.
by jebarker 2y ago
+1 this is the dream career!
424.
by jebarker 2y ago
Ah, I see. Yes, I agree.
425.
by jebarker 2y ago
Did you mean to reply to me? I don't see how this counters what I said. If you mean to imply that the questions use of statistics is a comparable problem to the current "government experiment" then I don't agree.
426.
by jebarker 2y ago
> It is not evil to ask these questions or to experiment with government I don't think anyone has a problem with the question being asked. It's the non-scientific method of experimentation that is troubling people.
427.
by jebarker 2y ago
> Research math takes a lot of perseverance. Yup, I think stubbornness/perseverance is the most useful transferrable skill I got from doing research math. It's a double-edged sword though as I often just can't give up work
428.
by jebarker 2y ago
The article isn't about AI tools. He's saying that people should consider becoming engineers working on building AI training systems.
429.
by jebarker 2y ago
My experience from proving a moderately complicated result in my PhD was that it's neither. There wasn't enough time to brute force by trying many complete solutions, but it also wasn't a single flash of insight. It was more
430.
by jebarker 2y ago
Why would people give the most significant political events of this century so far "a rest"?
431.
by jebarker 2y ago
> I've found that writing on the screen makes me prefer paper; reading on the screen makes me prefer books. I had the same experience with Remarkable. I've found I'm much happier now that I bought a color laser printer and
432.
by jebarker 2y ago
Sorry, I misunderstood the parent. I thought they were questioning why someone would wear these shoes for training.
433.
by jebarker 2y ago
> I don't understand why you would in the first place, what would be the reasoning here? I don't know for sure but if the shoes relieve stress on some parts of the body, e.g. leg joints or muscles, you can get more time trainin
434.
by jebarker 2y ago
Same - I'm 43 and I have very few memories of my childhood. My 20s are pretty much a blur now too.
435.
by jebarker 2y ago
> I think that complex solutions CAN be grasped by the large majority of voters Not sure I agree. Not because the large majority aren't intellectually capable, but because there isn't the interest/attention span in most pe
436.
by jebarker 2y ago
AlphaFold 2. It's not an LLM but it's not that different technologically. EDIT: that last sentence is actually a bit unfair to AlphaFold 2 as I consider it to be much more impressive technology than LLMs, but it comes from a commo
437.
by jebarker 2y ago
> it's being eroded away In the same way an axe erodes a tree
438.
by jebarker 2y ago
So it's not really LLMs you want to see fail, it's the businesses people are building around them. I can sympathize with that for sure but to me it seems obvious that there's many potentially beneficial applications of the te
439.
by jebarker 2y ago
Why is that your goal?
440.
by jebarker 2y ago
Someone first needs to design a rule set for a game that only permits the correct use of language but encompasses the entire breadth of language use. Then it's plausible. Thankfully for mathematics and code this seems plausible due to
441.
by jebarker 2y ago
People that use xi should lose their math credentials
442.
by jebarker 2y ago
s/LLMs/VLMs/g
443.
by jebarker 2y ago
You're correct but this isn't a good counterargument. We don't have the opportunity to change how human brains work, so we have to accept them as they are. That's not true of LLMs and there's no necessity that we de
444.
by jebarker 2y ago
i'm not actually sure whether they're special tokens in the sense of being in the vocabulary
445.
by jebarker 2y ago
The same idea is in both the R1 and S1 papers (<think> tokens are used similarly). Basically they're using special tokens to mark in the prompt where the LLM should think more/revise the previous response. This can be repeat
446.
by jebarker 2y ago
S1 (and R1 tbh) has a bad smell to me or at least points towards an inefficiency. It's incredible that a tiny number of samples and some inserted <wait> tokens can have such a huge effect on model behavior. I bet that we'll
447.
by jebarker 2y ago
Assuming the company stock has an upward trajectory the correct ratio to cite here depends on whether you're looking at the original grant or the value upon vesting. You could plausibly have grants that are 1/3 of your compensatio
448.
by jebarker 2y ago
That makes a huge difference to the desirability of taking the deal I'd imagine.
449.
by jebarker 2y ago
What typically happens to unvested stock grants with these severance packages?
450.
by jebarker 2y ago
> But the pictures are what make people click I wonder if that's a conscious or sub-conscious decision on the clickers part? It's hard to imagine anyone is consciously motivated to click on an article because of an AI generated

More ›