• ZebralLogic for evaluating LLMs (Was: Alan Kay's Dynabook fueled by Pro

    From Mild Shock@21:1/5 to Mild Shock on Thu Aug 1 18:24:25 2024
    Hi,

    A Prolog appendix could give a nice boost
    to your Edge Device Artificial Intelligence, possibly
    not having much problems with Zebra puzzles:

    A week ago, I posted that I was cooking a
    logical reasoning benchmark as a side project.
    Now it's finally ready! Introducing πŸ¦“ π™•π™šπ™—π™§π™–π™‡π™€π™œπ™žπ™˜,
    designed for evaluating LLMs with Logic Puzzles. https://x.com/billyuchenlin/status/1814254565128335705

    LoL

    Bye

    Mild Shock schrieb:
    Hi,

    The paper mentions: "Huge computing power
    of our modern laptops". If I look at my
    new iPad Pro M4 2024, I would say

    "Huge computing power of tablets", measurement
    have shown it is almost twice as fast as my
    laptops form ca. 2020. So lets do the following:

    Bring LPTP to Dogelog Player?

    Bye

    P.S.: This would give a new spin of Alan
    Key's vision of Dynabook. Can we run the
    Dynabook idea on Prolog?

    Joe Armstrong interviews Alan Kay
    https://www.youtube.com/watch?v=fhOHn9TClXY

    Mild Shock schrieb:
    Hi,

    I remember Robert StΓ€rk's disappearing from
    academic life at ETH Zurich all of a sudden.
    Did Ulrich Neumerkel now also disappeared not

    because the Scryer Prolog disaster, but after
    he figured out that failure slices are not hip
    enought? What could be more hip, are the modalities

    of Robert StΓ€rk's logic more hip now and even useful?

    Automated Theorem Proving for Prolog Verification
    Fred Mesnard etc.. May 2024
    https://lim.univ-reunion.fr/staff/fred/Publications/24-MesnardMP-slides.pdf >>

    Disclaimer: I am not deep into this theory,
    it has some ingredients that were floating around
    the 80's / 80's, not only in the millieau of ETH Zurich,

    but also in the vincinity of Gehard Jaeger, Bern.
    There are many alternative formalizations that
    can express termination etc.. But maybe LPTP is

    especially suited for Prolog?


    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)