drdeca

30 ms·

421.
by drdeca 2y ago
The popular culture is less concerned about ageism than racism. (Also, like, there are clear biological differences by age, which are sometimes relevant, while the concept of race is a mistake. It seems understandable that people would be l
422.
by drdeca 2y ago
She’s on the board of swanbitcoin? Where can I confirm that information? I didn’t see her mentioned on the page where 6 people affiliated with the company are listed, but I don’t think that is a list of people on the board, so I don’t think
423.
by drdeca 2y ago
> You could train a model that just takes natural language and outputs only code, with no conversion. Those models would be less general, and less interesting to non Technical people, which is why we don’t see them as much (imo.) Right,
424.
by drdeca 2y ago
I don’t think the last one is right. Fully connected and CNN are part of “what neuron is connected to what neuron” (though in the case of CNN, a number of corresponding “neurons” have equal weights going to/from them ). Also, “activati
425.
by drdeca 2y ago
They may do that for that reason, and it might also happen to help them produce the right output code. My point is just that (aiui), for any transformer model, holding the network constant but increasing the length of the code that is the i
426.
by drdeca 2y ago
Hasn’t it been shown that the complexity class of a transformer outputting a single token is TC^0 ? So, you can’t really expect it to directly output what some input code would output without allowing it extra computation steps by letting i
427.
by drdeca 2y ago
If you hadn't said "Before wringing our hands too much about antisemitism or anti-Asian prejudice in universities," that might have been true. If you had just said "here is this data", then you would have "just
428.
by drdeca 2y ago
Hm, I’m not really following how this language works… what’s the basic idea of it?
429.
by drdeca 2y ago
von Neumann algebras are operator algebras. They are algebras whose elements are operators on some Hilbert space. They also satisfy some other conditions. What I just said is not meant as a definition.
430.
by drdeca 2y ago
And what of the person who hired a detective to look for the person? Or the hospitals that refused to return the bodies until they were done with them, after the family contacted them?
431.
by drdeca 2y ago
Ah, yeah, I guess that probably is true of transformers in practice. I was thinking about something which strictly takes in a sequence of tokens and outputs a (possibly 1-hot) probability distribution over all possible next tokens. Such a t
432.
by drdeca 2y ago
Could the argument be rescued by some additional assumptions? I agree with, and have previously also stated, the point you make there about “any non-auto-regressive model can be converted into an equivalent auto-regressive model by […]”, bu
433.
by drdeca 2y ago
AGI is something fairly specific, yes, but depending on what you mean by “understand”, I don’t think it necessarily needs to “understand”? To behave (for all practical purposes) as if it “understands” is good enough. For some senses of “un
434.
by drdeca 2y ago
Perhaps a benchmark could be a good approximate upper bound for something without being a good approximate lower bound for that thing?
435.
by drdeca 2y ago
Isn’t the gradient descent used, stochastic gradient descent? I think that could matter a little bit. Also, the base model when responding to base64 text, most of the time the next token is also part of the base64 text, right? So presumably
436.
by drdeca 2y ago
I find this analogy kind of confusing? Wouldn’t the analogous thing be to say that our DNA doesn’t understand, uh, how we are able to skateboard? But like, we generally don’t regard DNA as understanding anything, so that not unexpected. Whe
437.
by drdeca 2y ago
Only in a sense of “derivative work” which is rather broad, and which I don’t think copyright law restricts (though this still needs to be settled by the courts). To be a copyright violation, it doesn’t suffice that the one work had a cau
438.
by drdeca 2y ago
At least when the ambient vector space is more than 3-dimensional, yeah. Specifically, a hyperplane generally refers to something with codimension 1. (So, when the ambient vector space is finite-dimensional, the dimension of a hyperplane is
439.
by drdeca 2y ago
> increasingly implausible numbers of orthogonal spatial dimensions in order to render them ‘geometric’. Implausible how? “geometric” doesn’t mean “embeds nicely in 3D space”. What’s wrong with talking about the angle between two L^2 fun
440.
by drdeca 2y ago
Suppose they said, “I used to be part of this club that I liked, but they had a rule that if anyone ever missed the weekly Tuesday meeting, they were kicked out, with no exceptions. I really liked that club, but being kicked out was worth i
441.
by drdeca 2y ago
While I don’t think a full intellectual-property-communism (as you phrase it) would be best, but I do think something a bit closer to it than we currently have would likely be better. (Mostly reducing copyright lengths a decent bit, closer
442.
by drdeca 2y ago
[edit: found another comment which you replied to, and now I see that you already knew all this. Oops. Original comment follows] Maybe you know the thing I’m about to say and just consider it to be too pedantic, but: The question to ask of
443.
by drdeca 2y ago
I think one difference is that you are seeing things as defaulting to “not allowed to use the work for whatever purpose, and the only reason it is ok for people to look at it to learn from it, is because they have a human right to do so, wh
444.
by drdeca 2y ago
Saying that it “launders” only makes sense under the position you are claiming. So, it might be fine as a conclusion/claim, which I guess is how you’re using it, but it wouldn’t be good to use as part of an argument leading to your con
445.
by drdeca 2y ago
When a drug first stops being protected by a patent, in order for another company to begin making a generic version, they have to pay some up-front costs (of course). If the government requirements for starting a generic for it are particul
446.
by drdeca 2y ago
Why does it matter whether their caring about whether their prediction is correct, is done at the time that their prediction is about?
447.
by drdeca 2y ago
In what I was imagining, the income was from people paying to be considered for inclusion on the list. Reading the list is free.
448.
by drdeca 2y ago
I don’t think “general” means like, equally suited for all cases. Rather, being able to handle the things successfully at all, even after working past wrong intuitions, and taking much time, still counts.
449.
by drdeca 2y ago
What do you mean by “as though our will is free” when you say that “we must operate as though our will is free”?
450.
by drdeca 2y ago
A thought I had: what if there were a service which maintained a list of arxiv identifiers (or, maybe also allow doi identifiers, but only if some extra conditions are met) that met their standards for notability/interesting-ness/

More ›