Anthropic, SpaceX and OpenAI are not banks. (Also, we had the largest bank runs in American history three years ago. The ordinary American barely noticed.)
And these equity-investors, do they use their own money to buy the (presumably non-voting) stocks?
Cause if that's the case, I see no reason for a government bailout should things go south. Nobody's pension would be affected by some private investor losing money on a bad investment.
But if that's not the case, then someone somewhere along the chain is acting as a bank, subject to a vibe-driven run.
> these equity-investors, do they use their own money to buy the (presumably non-voting) stocks?
Yes [1].
> Nobody's pension would be affected by some private investor losing money on a bad investment
...pensions also invest in the stock market.
> if that's not the case, then someone somewhere along the chain is acting as a bank, subject to a vibe-driven run
You're confusing deeply unrelated concepts. Whether or not someone who loses money is politically sympathetic has nothing to do with whether they're at risk of a bank run.
I made no mention of anyone being politically sympathetic or otherwise. A private investor is _private_ and thus not subject to a government bailout. The argument for government bailouts used to be that "grandpa would lose his pension", I merely stated the terms that would make this non-applicable.
If pensions invest in the stock market, then they are de-facto acting as a bank. And last I checked, in the land of the free, you get to withdraw your 401k should you vibe with the decision to do so [please don't do this based on this post alone].
It’s an important difference. Pension funds direct their investments. 401(k)s are self directed. Again, these words have meaningful differences you’re ignoring.
I'm not a systems guy, but you're probably right. A mutex requires an atomic variable which triggers the same cache coherency mechanism under contention.
What Justine probably meant was an apples to oranges comparison: in userspace, you can sometimes add constraints that allow for a faster algorithm whereas the cpu generally has to assume the worst.
That being said, if you look through this thread you'll find similar issues with OP's phrasing. A fine engineer / hacker, but too much 4chan troll in my book.
> There is a theorem that (...) multiplication requires unbounded memory
What theorem is that?
The multiplication of any two integers below a certain size (called "words") fits in a "double word" and the naive multiplication algorithm needs to store the inputs, an accumulator and at most another temporary for a grand total of 6*word_size
Sure, you can technically "stream" carry-addition (which is obvious from the way adders are chained in ALU-101) and thus in a strict sense addition is O(1) memory but towards your final point:
> Theory is important, but engineering is also important.
In practice, addition requires unbounded memory as well (the inputs). And it's definitely compute-unbounded, if your inputs are unbounded.
I dislike the term "we muddle along". IEEE 754 has well specified error bars and cases, and so does all good data science. LLMs do not, or at least they do not expose them to the end user
So then, how exactly do we go about proving that the result of chaining prompts is within a controllable margin of error of the intended result? Because despite all the specs, numerical stability is the reason people don't write their own LAPACK.
The video is about a 2x1 link, which the author hopes to eventually scale up to 3x4 using 40 gig transceivers. I'd say thunderbolt is probably safe in the near future.
Just to put some context into what _never_ means here:
If a website offers me the choice between "accept cookies" and "more options", I'll manually edit the DOM to remove the popup from the offending website. Some sites disable scrolling while such a "We value your privacy" popup is shown, so I wrote a js bookmarklet to work around most common means of scroll hijacking.
Google is currently waging a war against adblockers, especially on youtube. I currently have a way around that too but should they start baking ads in the video bytes, I'll stop using youtube altogether (though I am willing to look the other way for content creators shouting out their curated sponsors).
There is simply no universe in which I pay for certain types of digital content, and while I can't stop the data collection that ultimately pays for it, I can at least make damn sure that it's unlawful.
With respect to Spain and sports, stadiums are littered with ads, players wear ads, the commentator stream itself has ads baked in and people buy tickets and tapas to watch the game live. If that's not enough, go fuck yourselves!
Github: https://github.com/luckyPipewrench
> Pipe-layin' mastermind, code slayer supreme. Typin' fire across the screen, layin' pipe through the night. Bustin' blockages and bugs with no mercy.
That's a nice niche you found (spoken from one heap fan to another) but I have to say I strongly disagree with your use of *roughly* twice as much
At best you were off by one but in the context of performance, you'd want to assign that extra to a super-root of ±inf in order to save log2n extra range checks per heap-up, no?
Ah, I meant that for a classic heap, it's convenient to assign h[0] to the limit of your goal function (e.g. -inf for a min heap) cause that way you can skip the while i>0 check and just do while h[i>>1] > min(h[i], h[i+1]), asserting that the condition is definitely false when i < 2
I guess that it's not as important when you're explicitly willing to pay the piper and run the full log2n iterations just for the sake of branch predictability.
Thanks for the algo though, before today I never would've thought about doing a branchless downheap with i = i<<1 + h[i<<1] != h[i]
The good news is that these folks seem to be in possession of a vibe-rator.
reply