eru

26 ms·

481.
by eru 2mo ago
Why do you think on the job learning will fare better than college?
482.
by eru 2mo ago
We have a lot of competing frontier or near-frontier AI companies. So far a lot of the productivity improvements seem to be delivered as customer surplus.
483.
by eru 2mo ago
There are other countries.
484.
by eru 2mo ago
You can 'vouch' for a comment, if you think it shouldn't be flagged.
485.
by eru 2mo ago
Yes, naively understood in his sense 'best' word means that you pick the word with the maximum score, instead of sampling from the distribution. That doesn't actually give you the 'best' text in any human sense of t
486.
by eru 2mo ago
To sample from the probability distribution you already need random numbers. If you get your random numbers from a cryptographic PRNG, then to notice the difference between that and 'real' random numbers even in theory, means you
487.
by eru 2mo ago
> [...] and go back and restructure sentences and paragraphs if we want. That's what LLMs in reasoning mode do, too, to the text they present to you.
488.
by eru 2mo ago
Yes, and we are still pretty early: AI is still advancing at breakneck speeds, and hardware is too.
489.
by eru 2mo ago
My very own weightloss method is a lot simpler, and doesn't require anything beyond a trivial model of human metabolism: Take your weight today as the start. Decide on your target weight. Target trajectory := max(target weight, start
490.
by eru 2mo ago
I'm not quite sure why or how estimating metabolism helps here? I mean, observing the weightloss / gain or lack thereof itself will tell you whether you need more of a calorie deficit or surplus. What does an independent metaboli
491.
by eru 2mo ago
(3) seems like a problem in its own right? Basing science, traditional or newfangled ML, on such small amounts of data looks pretty weak.
492.
by eru 2mo ago
Not really. Expectations do most of that work.
493.
by eru 2mo ago
> You see, the national identity number is, literally, the number on your national identity card, or in other words the serial number of the document; not the person. When the document changes, the number changes with it. In Singapore, y
494.
by eru 2mo ago
If you have proper tail call optimisation, then tail calls are GOTOs with params. Trampolines allow you to simulate that, even when your compiler / language doesn't handle tail calls properly.
495.
by eru 2mo ago
I can write numbers like three by just writing 3 in my code. When I want a named number I use a syntax like x = 3. Why should functions be any different? A language doesn't need different ways to name things for each type of thing.
496.
by eru 2mo ago
I even tell them to give me an adversarial review and tell me what's wrong with the approaches.
497.
by eru 2mo ago
Well, the dotcom bust wasn't all that bad for the wider economy. No financial crisis. A shallow recession (and even that could have been avoided.) Btw, the dotcom bust was real, but there was no dotcom bubble. Skeptics back then sai
498.
by eru 2mo ago
Not really. Central banks don't really control relative prices.
499.
by eru 2mo ago
You can read the original photoelectric effect paper at https://myweb.rz.uni-augsburg.de/~eckern/adp/history/einstei... It's a bit more involved than the one about Brownian motion, but it's still fa
500.
by eru 2mo ago
These crises are manufactured by the central banks. Compare and contrast how the dot-com bust did _not_ lead to global financial crises. Nor did Black Monday, nor the recent string of bank failures in the US. ('Manufactured' abov
501.
by eru 2mo ago
Why is anything going to be catastrophic? Companies can go bankrupt without catastrophes for the rest of us. Happens all the time.
502.
by eru 2mo ago
> I thought they'd prefer their own solutions, but they frequently preferred the other model's solutions. I suspect that's downstream from their sycophancy.
503.
by eru 2mo ago
Maybe not embarrassingly parallel, but can be surprisingly parallel.
504.
by eru 2mo ago
In German the distinction is also different from what English speakers sometimes make.
505.
by eru 2mo ago
Oh, the book is probably good and entertaining! It's just not necessary, if you paid attention in high school.
506.
by eru 2mo ago
Hence the placebo control.
507.
by eru 2mo ago
Compare 'Lambda: the ultimate GOTO' https://www2.cs.sfu.ca/CourseCentral/383/havens/pubs/lambda-... which also uses state machines as its motivating example.
508.
by eru 2mo ago
> E.g., I've seen some people brag about some configuration language being easy to mechanically analyse because it's not Turing-complete, while in fact it's at least PSPACE-hard to analyse. Isn't even just the questio
509.
by eru 2mo ago
> My point is that each of the material above (and i forgot to add the original Quantum Mechanics papers) were major watershed moments in science/mathematics and hence are not easy to read/understand. I must dissent. Einstein&
510.
by eru 2mo ago
Hmm, article is also AI written. Oh, well.

More ›