ogogmad

18 ms·

331.
by ogogmad 2y ago
"No rumour is true until it's categorically denied" -- Otto Von Bismarck. > There is a very low cost to denying lies. So people can just lie. See: Clinton and Lewinski, the Profumo affair, Russian troop buildup on the Ukra
332.
by ogogmad 2y ago
It's probably nothing, but one of the pivotal lemmas is (a|(a|(a|b))) = (a|b). Imagine replacing (a|b) with x to get (a|(a|x)) = x; then this identity can be read as "x reflected through 'a' twice gives x". It remin
333.
by ogogmad 2y ago
Privacy?
334.
by ogogmad 2y ago
I think he meant that as a rhetorical question. It's commentary because he thinks it's Onion-worthy.
335.
by ogogmad 2y ago
One use-case: They help with learning things quickly by having a chat and asking questions. And they never get tired or emotional. Tutoring 24/7. They also generate small code or scripts, as well as automate small things, when you&#x27
336.
by ogogmad 2y ago
Look at the docs it links to.
337.
by ogogmad 2y ago
You need to look at the docs.
338.
by ogogmad 2y ago
It can explain its solution. Point to relevant docs as well.
339.
by ogogmad 2y ago
> They can't go beyond the boundaries of their training set. TFA says they just did. That's what the ARC-AGI benchmark was supposed to test.
340.
by ogogmad 2y ago
The article showed that Dirac deltas could be defined WITHOUT distributions. You ignored the article when answering my question. The question is why distribution theory is a particularly good approach to notions like the Dirac delta.
341.
by ogogmad 2y ago
The Last Samurai but with regex. Regizashi.
342.
by ogogmad 2y ago
The following is just my opinion: Integration can be done with its own special arithmetic: Interval arithmetic. I base this suggestion on the fact that this is apparently the only way of automatically getting error bounds on integrals. It&#
343.
by ogogmad 2y ago
I think you can get a generalisation of autodiff using this idea of "nonstandard real numbers": You just need a computable field with infinitesimals in it. The Levi-Civita field looks especially convenient because it's real-c
344.
by ogogmad 2y ago
Making a computer game is very expensive and time-consuming. This technology might allow a 12 year old to produce a fully working AAA-quality game on their own for almost nothing. But sigh it's an early demo that needs some improving
345.
by ogogmad 2y ago
Maybe this is at cross-purposes: Functions don't have memory. Nothing real ever "is" a Markov chain. It's simply a (slightly unusual) modelling choice which you can use whenever you've listed all possible inputs tha
346.
by ogogmad 2y ago
The abstract mentions "stationary distributions". So they try to derive something new from that borderline obvious equivalence. The Google search engine originally used the stationary distribution to rank the importance of search
347.
by ogogmad 2y ago
They use O(T^K) states where T is the size of the "vocabulary" (which would in fact be tokens in the case of an LLM) and K is the size of the context window. In practice, this would be an enormous number of states, and would be ab
348.
by ogogmad 2y ago
This doesn't count because the idioms are very different from Go and Rust. And I suspect there's a high learning curve for features like Functors. If you check the Wikipedia page for OCaml to find out where it gets used, you'
349.
by ogogmad 2y ago
This approach makes sense if you have access to either all available memory or a fixed quantity of memory, half of which would go to the fromspace and the other half to the tospace. What I don't understand is: When should the heap be m
350.
by ogogmad 2y ago
The course doesn't mention Concentration Inequalities as far as I can tell. Are these perhaps under-used in stats? It seems the alternative is to guess probability distributions, which is less rigorous.
351.
by ogogmad 2y ago
This is a weirdly arrogant comment given both TFA and the fact that professional physicists have worked on MOND and disagree with everything you've said. https://www.youtube.com/watch?v=n33aurhg788 Is this typical beha
352.
by ogogmad 2y ago
You still need to load it.
353.
by ogogmad 2y ago
There's also adversarial images. I wonder if those can used here.
354.
by ogogmad 2y ago
You can use a screenshot to get around it. It's a case of the "analogue hole".
355.
by ogogmad 2y ago
Yes. Compare it to the List monad or Parsec.
356.
by ogogmad 2y ago
I imagine this could be understood as making use of a monad. Right? The main problem with generalised regexes is that you can't match them in linear time worst-case. I'm wondering if this is addressed at all by Raku.
357.
by ogogmad 2y ago
I've looked up alloca() and it's considered about as safe to use as recursion, and potentially safer than stack-allocating one big fixed-size array. Its bad reputation might be because of poor implementations in certain compilers.
358.
by ogogmad 2y ago
ALGOL is the main forerunner of C. Anything vaguely C-like is also ALGOL-like.
359.
by ogogmad 2y ago
I've just looked at this snippet, which the blog used back in 2017 to illustrate that Algol 68 was "utter madness": https://craftofcoding.wordpress.com/2017/03/06/a-brief-look-... COMMENT
360.
by ogogmad 2y ago
Oh yes, that's fair. I was thinking about that, but somehow dismissed it.

More ›