setr

37 ms·

571.
by setr 5y ago
Should GPU choice affect model performance? AFAIK it should only impact training time
572.
by setr 5y ago
It seems to me the argument is really that it’s meant to be varied — as in, their goal was really to be playing with the awkwardness of poorly constructed tooling that can only haphazardously be declared an “instrument”. Akin to the “hacker
573.
by setr 5y ago
I generally describe the problem as: knowledge is nodes in a graph, wisdom is making connections between nodes in that graph. Framing curiosity and smartness under that model: curiosity is your ability to add nodes, smart is your ability to
574.
by setr 5y ago
I don't think it matters whether its that statistically significant. You can increase expected-value by increasing likelihood of occurrence, or increase the value of the outcome. Take a loan shark, for example. You don't need to t
575.
by setr 5y ago
The first thing I try to do upon encountering any technology/stack people recommend everywhere is try to figure out what’s wrong with it. I then skim through anything I can find that’s relevant, and dig into the parts I didn’t fully un
576.
by setr 5y ago
Facebook does something productive, but I don’t see how stocks map to that productivity. The only way I really see to take a stock and have it represent actual value, is to plug it into the stock market. But this is just trading between tho
577.
by setr 5y ago
This skips the problem, again. You can trade a sheep as much as you want, between any numbers of investors. But there is some basic value to sheep, because eventually someone is able to convert the sheep into money, with effort time and pro
578.
by setr 5y ago
The part I don't understand is how do I translate "ownership in apple" to USD. Obviously I can trade it to someone else, but that's just moving the problem -- eventually someone has to be in the position to actually extr
579.
by setr 5y ago
This I've never understood -- if a stock is detached from company earnings, then what drives the value of a stock, beyond the meaningless "someone else is willing to pay for it" -- why are they willing? It's not at all c
580.
by setr 5y ago
JS would like to have a word with you.
581.
by setr 5y ago
given the cost difference between storage and bandwidth, that's probably a worthwhile tradeoff. Especially given the common usecase of store once, send many times.
582.
by setr 5y ago
So the same prey saturation strategy as cicadas? I wonder if that also produces a prime-number cycle preference
583.
by setr 5y ago
I imagine it’s like a b-tree. You simply have more reports, until it becomes untenable — it takes more time to coordinate them than you have. So your tree splits — you pick a couple nodes to continue directly reporting to you, and all other
584.
by setr 5y ago
It’s easy to show that something is incorrect, because it doesn’t meet the base requirements of the question posed, or correct because it does. It’s difficult to show a correct answer is good, or the best, for the question. Many answers you
585.
by setr 5y ago
Pathfinding is largely not that bad — hierarchal A* for long distance pathing, flow field for dense agent areas (eg your fort). With sufficient density, it’s worth recalculating the field repeatedly. Partitions can also be used for resource
586.
by setr 5y ago
I think that’s just more a result of the game style/architecture. I don’t think there’s much hidden state that’s not encoded into the map/entities, and even things like AI is I think ultimately preserved as a simple stack of pendi
587.
by setr 5y ago
It’s only illegal if you get caught.
588.
by setr 5y ago
I recall trying to look into barcode scanners for personal use (figured I could catalogue everything in my house) but found it surprisingly difficult to find recommendations that seemed reliable, fast and integrated with software usefully.
589.
by setr 5y ago
According to everything I can google, camels are definitely domesticated, and the vast majority of them are today. According to wikipedia, there’s technically only 1400 “wild” camels or so, from the wild Bactrian group Some other neat facts
590.
by setr 5y ago
It’s not like… that’s precisely the scenario being referred to. The problem is that’s always the example, which suggests that the dropbox event is likely closer to the exception than the rule. And it’s existence is used to deflect all cri
591.
by setr 5y ago
Don’t know about “optimal”, but apparently the front crawl itself isn’t too novel.. from that wiki page: > The "front crawl" style has been in use since ancient times. There is an Egyptian bas relief piece dating to 2000 BCE sh
592.
by setr 5y ago
Pretty much all innovation large companies do is through acquisition. It’s much faster and risk-free than doing it yourself, from scratch.
593.
by setr 5y ago
The money quote: > The findings highlight the variety of ways the multiple father figures may have been helping the mom support her baby, whether it was providing food, housing, childcare or financial benefits," Bullock said. "
594.
by setr 5y ago
They need to be google-friendly, or they need to exist in a world where code-friendly search tools exist.
595.
by setr 5y ago
Doesn’t the three-letters do that like… all the time? That’s basically the definition of honeypotting As I recall, the FBI did exactly that fairly recently, when they pwned some credit card trading site and ran it themselves for years.
596.
by setr 5y ago
Right, but the game needs to actually build with it in mind. In a CRPG, death is basically an illegal operation — your only option is to save scum or reset the campaign (eg on TPK. On single death, you can probably resurrect), so the tail r
597.
by setr 5y ago
It’s a hierarchy… so you can take your pick of granularity? Only thing that matters is min, max and step-size. Unless we’re talking about different things, I’m not clear why we’re not assuming the tool you remember chose h10 or whatever lev
598.
by setr 5y ago
https://h3geo.org/docs/core-library/restable The smallest granularity is . 0000009 km^2, so I guess 1/3 of an inch. Which seems to me a ridiculously and unnecessarily precise, but who knows
599.
by setr 5y ago
I was thinking more like LoL, which as I recall implements a dynamic crit system — as crits occur, your % chance of a crit drops (possibly below your stated %), and as crits fail, your % chance of a crit increases (possibly above). Your eff
600.
by setr 5y ago
The problem is accounting for the 1-25, 1-125, 1-625 possibilities that are still legal, but you can’t meaningfully do anything about (and a 1-625 crit sequence trivializes any boss fight, and a 1-625 miss sequence makes a trivial enemy imp

More ›