sfink

33 ms·

451.
by sfink 2y ago
If you're going to check then you shouldn't be saturating, you should just be checking.
452.
by sfink 2y ago
> For example, most doctors believe in vaccines, but that's not group think, it's based in evidence that they have studied. From the outside, it might seem like group think. I'm willing to bet that in most cases, that is
453.
by sfink 2y ago
Well, "three months old" does sort of imply "zero years and three months". But thank you for reminding me that I am zero centuries old. More decades old than I would like, but zero centuries.
454.
by sfink 2y ago
I believe the main argument (from the OP) is that you have to specify the range with two bounds, and that it is common to want a 0 (assuming a 0-based indexing world), and so in order to refer to a range that includes index 0 you'll ne
455.
by sfink 2y ago
Values take up space. When you manipulate a value or pass it around, it makes no sense to sometimes refer to the beginning of the value and sometimes to the end. It makes a little more sense if you have an array of something other than plai
456.
by sfink 2y ago
When I saw the title, I knew what this was going to be. It made me want to immediately write a corresponding "Human Blindspots" blog post to counteract it, because I knew it was going to be the usual drivel about how the LLMs unde
457.
by sfink 2y ago
In the same FAQ: > Don’t Mozilla products, like Firefox, earn income? > Firefox is maintained by the Mozilla Corporation, a wholly-owned subsidiary of the Mozilla Foundation. While Firefox does produce revenue — chiefly through search
458.
by sfink 2y ago
The government should stop wasting money on things that I don't care about. (Especially the regulatory things that apply to me personally.)
459.
by sfink 2y ago
This is a good idea on paper. But it turns out that having a for-profit corporation (as Mozilla Corporation is) accept donations, especially donations earmarked for certain purposes, is understandably tricky from a regulatory and tax standp
460.
by sfink 2y ago
Don't forget that this is driven by present-day AI. Which means people will assume that it's checking for fraud and incorrect logic, when actually it's checking for self-consistency and consistency with training data. So it s
461.
by sfink 2y ago
If I'm understanding correctly, this will break whenever the IP address of your tunnel changes. You'll have to reestablish all of your connections. My use case for tailscale: have an SSH (or other) connection to my home server whi
462.
by sfink 2y ago
Huh. If I had to pick a constraint to jettison from that list, I think I'd pick 1 == 1.0. In fact, I'd prefer an error when testing 1.0 == 1.0. It only makes sense to compare floats in very limited situations. While admittedly thi
463.
by sfink 2y ago
Because there is a high barrier to entry. In the US, at least, there was also an explicit policy and set of mechanisms to limit supply of doctors: https://www.advisory.com/daily-briefing/2022/02/16/physic
464.
by sfink 2y ago
That's ok. Writing such a spec is writing the code, declaratively. The only difference between that and writing SQL (as opposed to writing imperative code to query the database) is that the translation mechanism is much more sophisti
465.
by sfink 2y ago
I don't care about the programmers who can't write FizzBuzz. Why should I? If I employed them, they were costing me money. If I worked with them, they were costing me time and hair follicles. I need them about as much as I need a
466.
by sfink 2y ago
Then you'll get code that passes the tests you generate, where "tests" includes whatever you feed the fuzzer to detect problems. (Just crashes? Timeouts? Comparison with a gold standard?) Sorry, I'm failing to see your p
467.
by sfink 2y ago
Of course I care. Human-written code was written for a purpose, with a set of constraints in mind, and other related code will have been written for the same or a complementary purpose and set of constraints. There is intention in the code.
468.
by sfink 2y ago
> I do worry about model poisoning with fake truths but don't feel we are there yet. In my use, hallucinations will need to be a lot lower before we get there, because I already can't trust anything an LLM says so I don't
469.
by sfink 2y ago
> feel free to copy and claim as your own. That's a nice sentiment, but I'd encourage you to add a license or something. The basic "something" would be adding a canonical URL into the spreadsheet itself somewhere, alo
470.
by sfink 2y ago
That's what you'd think, but actually violence mostly pushes things to the extreme and sinks the effort permanently. Nonviolent movements are often defeated by one side goading the other into losing control and doing something vio
471.
by sfink 2y ago
My suspicion is that this is not answerable in general. To a first approximation: whichever of the two your political leanings make you more comfortable with. To a second approximation: if your task is efficiency at a constrained task, homo
472.
by sfink 2y ago
One tiny missing bit: drawing with XOR wasn't just to be able to easily recover the previous state. If you're on a black and white display and are dragging around a rectangular outline, then dragging it over a filled portion of th
473.
by sfink 2y ago
Exactly. During minor GC, you update the promotion rate of each callsite. If it gets too high, you start allocating things from that callsite directly from the tenured heap, since your statistics are telling you that the things allocated at
474.
by sfink 2y ago
My understanding is that splay is more of a benchmark that you try to make generational GC not hurt too much. It allocates lots of long-lived objects that it holds onto for the full test, and for JS at least it then hangs strings off of tho
475.
by sfink 2y ago
I agree that the commentary is nice, and should be paid attention to, but it's also from 2007. So perhaps you mean to say "it would be useful to consider this paper in the context of this 2007 commentary"? I would agree with
476.
by sfink 2y ago
The cat's out of the bag, if someone stops then someone else would start. I would entreat people to consider the net effect of anything they create. Let it at least sway your decisions somewhat. It probably won't be enough to not
477.
by sfink 2y ago
I agree, but LLMs in general have a horrendously bad smell in terms of efficiency. s1 and r1 are just proving it. The models' latent spaces are insanely large. The vast, vast majority pretty much has to be irrelevant and useless, it&#x
478.
by sfink 2y ago
Sorry, I knew I remembered reading about the two somewhere but could not figure out where. I searched through the jj page for it. Only later did I remember that it came from your jj guide, which I still had open in a tab. I wish I had point
479.
by sfink 2y ago
I've been using jj for a month, coming from mercurial. I have settled into a usage pattern slightly different than the article describes, so that all of those `jj edit` commands made me a little itchy. I prefer to nearly always use a s
480.
by sfink 2y ago
Yes. That is what "...unless the base rate is set externally somehow" means. It is different initially, when only one company is offering the "discount" and they have not yet adjusted their base price upward. In fact,

More ›