YeGoblynQueenne

621 ms·

1. 2611.▲
   by YeGoblynQueenne 4y ago

   >> Back in 541-542 an outbreak killed about 40% of the city’s population. An outbreak of the plague .
2. 2612.▲
   by YeGoblynQueenne 4y ago

   That's because Byzantium was the eastern wing of the Roman Empire. Justinian I (the emperor in the wikipedia article) is remembered for being behind the last almost successful attempt to take Rome back from the hands of the barbarians
3. 2613.▲
   by YeGoblynQueenne 4y ago

   Oooh, I remember now. You are talking about that discussion we had a while ago. OK. Yeah, so,"learning" in that context means you're given a bunch of atoms as examples and you learn a logic programs that entails those atoms
4. 2614.▲
   by YeGoblynQueenne 4y ago

   I don't know that Prolog, or logic programming, makes any promises about formal semantics. I have to say I don't think of Prolog or logic programming in that context. Except for the more general point that Prolog programs are theo
5. 2615.▲
   by YeGoblynQueenne 4y ago

   Like I say above, that's the difference between using a green cut vs. a red cut. See also my link to the RLE version in "99 Prolog Problems". So far it's been my experience that it's very hard to find a Prolog progr
6. 2616.▲
   by YeGoblynQueenne 4y ago

   I don't remember having any problems with debugging programs with cuts. But it should be said that I restrict myself to green cuts [1]. There are usual ways to avoid those, but not without making the code more complicated than it needs
7. 2617.▲
   by YeGoblynQueenne 4y ago

   Hey, JerremyBanks [dead], check out my comment history. Should help answer your question ;)
8. 2618.▲
   by YeGoblynQueenne 4y ago

   Less than 5 years to AGI? An "AGI" that has to be retrained from scratch, with more data and more compute, and with a modified architecture, every time we want to improve its performance? An "AGI" that forgets every inte
9. 2619.▲
   by YeGoblynQueenne 4y ago

   >> It can absolutely prioritize and plan, you literally just have to ask it and loop on itself. GPT-4 can't perform simple classical planning tasks so it's doubtful that it can "plan" in a looser way, as I think yo
10. 2620.▲
    The implausibility of intelligence explosion (2017)(medium.com)

    5 points by YeGoblynQueenne 4y ago|0 comments
11. 2621.▲
    by YeGoblynQueenne 4y ago

    >> PowerShell has its own alias, "gci" for get-childitem and just "childitem" would automagically work. Yeah, I totally keep forgetting those and just use the unix-like aliases. >> btw. what is the benefit of
12. 2622.▲
    by YeGoblynQueenne 4y ago

    Windows Experience Points? Sorry, it's all tongue-in-cheek.
13. 2623.▲
    by YeGoblynQueenne 4y ago

    Prolog doesn't have typed variables, but any Prolog term has a type, the set of Prolog terms that unify with itself. In practice, when you write Prolog, your arguments to predicates are always implicitly typed, for example if you write
14. 2624.▲
    by YeGoblynQueenne 4y ago

    Woa, blast from the past! :D I used Greibach normal form grammars in my Master's, in 2014, when I was doing some grammar induction stuff with DCGs. Now, Greibach normal form is useful, but its use is to avoid left-recursions. That&#x27
15. 2625.▲
    by YeGoblynQueenne 4y ago

    >> Just try writing (without hacks/cuts) a run-length encoder that produces only one answer ([9,9,9,8,8,7] -> [[9,3], [8,2], [7,1]]) and works forwards and backwards and then tell me the execution model is not a limitation. Bu
16. 2626.▲
    by YeGoblynQueenne 4y ago

    I think what you're saying is that you can take a Prolog program and ground it all down to the propositional order so you can then evaluate it with a SAT solver. For example, you can take a Prolog clause like this one: p(X,Y):- q(X
17. 2627.▲
    by YeGoblynQueenne 4y ago

    Prolog is not first-order. That's how it's described but it lacks the fundamental distinction between predicate and function symbols that makes First-Order Logic, first-order. To clarify. A FOL language begins with an "alphab
18. 2628.▲
    by YeGoblynQueenne 4y ago

    >> Case in point: basically all your mentioned alternatives (Datalog, SMT, I'd add ASP) have implementations that use the Prolog syntax or something close to it with very similar semantics. But not being tied to that specific exe
19. 2629.▲
    by YeGoblynQueenne 4y ago

    Prolog is not a DSL. Or I guess you can see it as a DSL where the "Domain" is the entire set of programs computable by a Universal Turing Machine. It's a DSL for writing programs that manipulate programs that manipulate progr
20. 2630.▲
    by YeGoblynQueenne 4y ago

    Besides, it's just faster to write "ls" than "Get-ChildItem". Linux command names come from a time when pressing keys on a keyboard was very expensive [1], so they're very economical. Nowadays we can type keyb
21. 2631.▲
    by YeGoblynQueenne 4y ago

    Sorry, I can't edit my earlier comment but I re-read our conversation and I realise I misread your quote. In particular: >> read a product description and generate the appropriate codebase Chollet here is talking about creating a
22. 2632.▲
    by YeGoblynQueenne 4y ago

    Sorry, Isekai fans, or vending machines?
23. 2633.▲
    by YeGoblynQueenne 4y ago

    (Not the OP) What I wonder about is why would performance increase with what you suggest. In my mind, you're proposing getting a language model to generate its own training data. But data on its own is not enough to train a good model.
24. 2634.▲
    by YeGoblynQueenne 4y ago

    Thanks for the paper, I need to read that. Much of my work is on inductive logic. >> I think that Penrose basically thinks it's useful that humans can heuristically step up from a logic system S to S+"statement X is true&quo
25. 2635.▲
    by YeGoblynQueenne 4y ago

    >> Putting aside any thoughts of whether this is persuasive, I think this seems to be a case of Lucas and Penrose doing what they claim is uniquely human: thinking outside of the box of formal proof. More generally, almost none of hum
26. 2636.▲
    by YeGoblynQueenne 4y ago

    Thanks for clarifying! Sorry for pouncing on that one little detail in your comment btw. Incompleteness vs. human reasoning is just something I think about on and off and your turn of phrase caught my attention.
27. 2637.▲
    by YeGoblynQueenne 4y ago

    I'm sorry, I'm not familiar with the grokking results. What are you referring to?
28. 2638.▲
    by YeGoblynQueenne 4y ago

    I wonder too. I've communicated with Chollet in the past (about his ARC dataset) and he's responsive. You could email him and ask, or you could ask him on his twitter (I don't use twitter).
29. 2639.▲
    by YeGoblynQueenne 4y ago

    >> If a model truly generalizes, why wouldn't it be possible for it to generate more real signal than was used for training it? If by "a model [that] truly generalizes" you mean one that generalises outside its training
30. 2640.▲
    by YeGoblynQueenne 4y ago

    (...) Gödel-unprovable results are provable by humans. Are they? Certainly humans can recognise the undedicability of a "barber's paradox", like "this sentence is false", but that doesn't mean we can actually d