sfink

35 ms·

481.
by sfink 2y ago
That "better" analogy is a restatement of "$5 for lemonade, $3 off if you skip the lemon." > In this case, the discount is "opt-in." The base price is not a force of nature. $5 with the option to opt-in to a
482.
by sfink 2y ago
Some form of this is quite common. My town's water treatment plant uses fish. They have all of their automatic monitoring things, but they also have a school of fish in the water coming out of the last stage of the plant, just before r
483.
by sfink 2y ago
Time switches can only roughly accommodate the changing time of dusk and dawn, and can't handle dark weather at all. Said otherwise: if you're trying to approximate a light sensor, a clock is a worse approximation than a live chic
484.
by sfink 2y ago
> Do some insurance companies charge less when provided access to policy owner driving patterns which the companies infer reduce their risk? Sure. > But that is a different question. In what way? A discount for allowing surveillance i
485.
by sfink 2y ago
I'm not going to buy either, so perhaps my opinion here doesn't count. But I'd be far more likely to buy a set, one for me and one for my wife, than I would be to just buy one for myself. I think that's why she made it,
486.
by sfink 2y ago
Interesting. In my dealings with younger (and older) kids, thinking of them as little people seems to work well. Little, fully complete people, with many parts dramatically simpler (and more straightforward) than adults. But then, I also th
487.
by sfink 2y ago
Sure, but don't lose sight of the fact that female insecurity has been leveraged and marketed to as well, for pretty much forever. Fashion, makeup, beauty standards -- flip through a magazine and look at the ads, and think about what t
488.
by sfink 2y ago
Oh. Er... I didn't really consider whether it could be proven to have an integral, I just went off of intuition. After an additional lifetime of study, magically compressed into 15 minutes of skimming Wikipedia articles, I'm going
489.
by sfink 2y ago
I'll throw in my funky function (which is probably similar to a well-known one, but I'm no mathematician so I know not what is well known): f(x) = a random number uniformly selected from [-1, 1]. (So every real number maps to a di
490.
by sfink 2y ago
Hm, it doesn't seem problematic for the area to be lower than expected for a given perimeter. You could do the same thing by having a triangle (an actual triangle), but then at some random point have the bounding line shoot off in some
491.
by sfink 2y ago
Please be very, very careful with security. That includes logging as little as you can get away with. Your site is a tempting target for a certain class of people.
492.
by sfink 2y ago
I think you're missing the point. The comparison is not between 93% and 92%. The comparison is between what they're getting (93%) and what you'd get if you scaled up the usual process to the core size they're using (0%).
493.
by sfink 2y ago
> That requires a high trust environment; I don't know if it could work for Firefox where you may have some very part-time contributors. Ironically, that's why it works for Firefox. Contributors follow a power law. There are
494.
by sfink 2y ago
Hopefully, you have a monorepo or something with similar effects, and a lack of fiefdoms. In that case, if the current way is undocumented and/or inconsistent, you make it better before or while adding in your new approach. If there ar
495.
by sfink 2y ago
Personally, I think that part is just incomplete. I think that there is an attitude that because you have jumped into several older open source projects and successfully made meaningful changes, that you have mastered the art of diving into
496.
by sfink 2y ago
Ah, I understand your logic now. Fair enough, and since I'm not benchmarking LLMs to begin with, I guess it doesn't much matter whether I fully agree or not. But: > You can always run the new question retroactively on an old mo
497.
by sfink 2y ago
Not any that you can think of. Just the ones you've published something about. Plus, simonw isn't exactly a meaningless nobody in this space, and his writeups are more detailed and actionable, and therefore identifiable, than some
498.
by sfink 2y ago
I'm somewhat this way, but more than somewhat when it comes to TV shows. I think something like 80% of my favorite shows were cancelled early. Firefly, Freaks & Geeks, Shadow & Bone, ... Perhaps I should set up a Patreon where
499.
by sfink 2y ago
I hope you have other private benchmarks running that you don't talk about or publish, just in case a model maker intentionally targets one of your benchmarks, or some fuzzy "find things people have mentioned as potential LLM benc
500.
by sfink 2y ago
No, it means the static analyzers can't report on a different error because a subset of that class of errors is no longer an error, and the static analysis can't usually distinguish between that subset and the rest. memcpy
501.
by sfink 2y ago
> Similarly, GCC may delete a memcpy to a buffer about to be freed, although I have never observed that as you generally don’t do that in production code. It's not that crazy. You could have a refcounted object that poisons itself w
502.
by sfink 2y ago
The toxicity example was thought-provoking. > Input: Caspar Weinberger's father, Herman, was the younger > Model generation for input: son of an immigrant grocer. > Perspective API on model generation: Toxic I
503.
by sfink 2y ago
You're just renaming "error" to "uncertainty". That is incorrect. Everything unwanted is error, by definition. All of the heuristics are about reducing error, because that's what the goal is. Some of that error
504.
by sfink 2y ago
If it works properly, it would need to say that it doesn't know that it doesn't know, and then where are you? (Short answer is yes, but it only works for a limited set of things, and that set can be expanded with effort but will a
505.
by sfink 2y ago
> proving that hallucination is conceptually a simple problem. ...proving that this one particular piece of the hallucination problem may be conceptually simple. FTFY
506.
by sfink 2y ago
The term is actually fine. The problem is when it's divorced from the reality of: > in some sense, hallucination is all LLMs do. They are dream machines. If you understand that, then the term "hallucination" makes perfect
507.
by sfink 2y ago
I tried this, but it didn't work very well for me. Maybe I was just catching people at a bad time? They'd always brush me off with some version of "uh, it isn't my birthday" "you've said that to me
508.
by sfink 2y ago
It was illustrating narcissistic personality disorder. Debating whether or not Musk delivers, or what personality traits might be behind delivering or not, is irrelevant to whether or not Musk illustrates NPD. It is easy to find examples of
509.
by sfink 2y ago
If you care about the experience of your co-workers, you won't repeatedly break stuff. (Or if you do, it's because your tooling hasn't scaled enough to keep up; you need a staging area or precommit hooks or whatever.) If you
510.
by sfink 2y ago
I don't see where your assumption is being challenged. "Breakfast cereals" is a very loose category. I would not conclude that Froot Loops lower risks of cardiovascular disease just because Ezekiel's Gravel Bits have bee

More ›