thesz

22 ms·

391.
by thesz 2y ago
> trained the model on synthetic data You get knowledge collapse [1] this way. [1] https://arxiv.org/abs/2404.03502
392.
by thesz 2y ago
This thing is awesome! I tried it on couple of one liners and it found a couple of potential problematic points, one for each one liner.
393.
by thesz 2y ago
> Humans really haven't changed that much at all. I guess you are quoting Woland from The Master and Margarita [1], the words he said in a show at the Variety theater. [1] https://en.wikipedia.org/wiki/T
394.
by thesz 2y ago
Why are there no European, American and/or Chinese vehicles to compare to? https://www.carpro.com/blog/almost-all-new-vehicles-have-aut... Why only those three?
395.
by thesz 2y ago
>> Recognizable repetition can be abstracted > ... which is the whole idea behind training, isn't it? The comment I was answering specifically dismissed LLM's inability to answer same question with same... a
396.
by thesz 2y ago
> And i have not seen a single customer demonstrate that they care about the quality of the code base behind the product they enjoy paying for. The code quality translates to speed of introduction of changes, fixes of defects and amount
397.
by thesz 2y ago
Recognizable repetition can be abstracted, reducing code base and its (running) support cost. The question that matters is: will businesses crumble due to overproduction of same (or lower) quality code sooner or later.
398.
by thesz 2y ago
Adds are cheaper only for fixed-point computations. Floating point addition needs to denormalize one of its' arguments, perform an (integer) addition and then normalize the result. Usually FP adds take a cycle or two longer than FP mul
399.
by thesz 2y ago
First, surrounding your code is other code that very probably have a clue on what you should do. You do not have to go at the definition to see what to do. Second, for one written/changed line I have read, perhaps, couple of hundredths
400.
by thesz 2y ago
As you mentioned machine learning, a useful way to implement inference in language models is to use a beam search: https://en.wikipedia.org/wiki/Beam_search Beam search approximates NP-hard solutions (make wider beam,
401.
by thesz 2y ago
Turing published the same year as Church, couple of months later. Church's advantage was very small. Turing machine includes write over some place, thus it models a calculating man with infinite paper, infinite pencil and infinite eras
402.
by thesz 2y ago
To me it was eye opening on how skewed is the presentation of these quantum computing number factoring advances. It is completely in line of the "The Case Against..."
403.
by thesz 2y ago
https://eprint.iacr.org/2015/1018.pdf As pointed out in [57], there has never been a genuine implementation of Shor’s algorithm. The only numbers ever to have been factored by that type of algorithm are 15 and 21,
404.
by thesz 2y ago
Yes, it contains "I) Update (Dec. 10): The Wind in the Willow"
405.
by thesz 2y ago
I think you will be pleasantly amused by this marriage of logic and religious beliefs: https://en.wikipedia.org/wiki/G%C3%B6del%27s_ontological_pro... It is formally proven that an entity that encompass all qualities o
406.
by thesz 2y ago
> maybe the whole project should have been scrapped at the planning phase. This is an argument made by one of the Dostoevsky characters, a famous "tear of a child" argument. [1] https://en.wikipedia.org/wiki&#x2
407.
by thesz 2y ago
There are several problems with the paper and with the idea itself. First, code is unavailable - github repository [1] does not contain any code. [1] https://github.com/tgchomia/ts The algorithm, itself, has some black
408.
by thesz 2y ago
> There needs to be a way to write a query that runs on an abstract interface, rather than an actual table. Proper use of SQL inverts control. Instead of parameterizing query by table, you write a query and at the actual use site you joi
409.
by thesz 2y ago
> IQ is Gaussian. The Big Five Personality Trait known as Conscientiousness is likewise Gaussian. For what it’s worth, human height is also Gaussian... Height cannot be negative, thus, it is not Gaussian. IQ cannot be negative too. Great
410.
by thesz 2y ago
Python optimizations are not deep and/or broad. I did not mean "very fast" as in "compilation speed." Python does not attempt to apply anything more complex than that peephole optimizer , whatever that means. Judgi
411.
by thesz 2y ago
Worst-case optimal join algorithm [1] might have a different opinion. [1] https://en.wikipedia.org/wiki/Worst-case_optimal_join_algori...
412.
by thesz 2y ago
Python does not do optimizations at all. In contrast, you can find Fortran compilers that can substitute your inefficient implementation of some numerical algorithm with different much more performant one.
413.
by thesz 2y ago
If axioms, facts and rules have types, these types can guide choice.
414.
by thesz 2y ago
You are building an SQL in disguise. First, you need to encode "memes" and relations between them at scale. This is not a language problem, it is data handling problem. Second, at some point of time you will need to query memes an
415.
by thesz 2y ago
So, you looking for Cyc [1], practically. [1] https://en.wikipedia.org/wiki/Cyc
416.
by thesz 2y ago
So you are directly express context-sensitivity instead of approximating it with infinite context-free grammar. Okay, good enough.
417.
by thesz 2y ago
The question comes from Haskell, yes: https://byorgey.wordpress.com/2012/01/05/parsing-context-sen... You used monadic parser, monadic parsers are known to be able to parse context-sensitive grammars. But, th
418.
by thesz 2y ago
How can one define an infinite grammar in Rust? E.g., a context-free rule S ::= abc|aabbcc|aaabbbccc|... can effectively parse a^Nb^Nc^N which is an example of context-sensitive grammar. This is a simple example, but something like that can
419.
by thesz 2y ago
In Agda, if_then_else_ is a ternary operator: https://wiki.portal.chalmers.se/agda/ReferenceManual/Mixfix For Agda being lazily evaluated, ternary if_then_else_ corresponds to usual if cond then true_expr else el
420.
by thesz 2y ago
Zero-g adds nothing. You cannot even purify silicon in zero-g, you need some-g for impurities to go up and down. The average distance between molecules of the atmosphere is 3.3nm, this is about 10 times of the typical atom diameter. You nee

More ›