bjourne

45 ms·

931.
by bjourne 2y ago
They have exhausted all other options and direct action is all that remains. They aren't misguided, you are.
932.
by bjourne 2y ago
Flight is an easy target because alternatives exist or could with relatively little effort be made practical. High-speed rail is a thing. And it's not like anyone actually needs to fly to South-east Asia for vacation every year.
933.
by bjourne 2y ago
> It is much more frequent for -ffast-math to generate bigger errors than to benefit from an accidental cancellation of the errors. In my experience underflows are very uncommon and indicate that you're doing something wrong, wherea
934.
by bjourne 2y ago
It's not so easy because -ffast-math can actually be more correct than running without.
935.
by bjourne 2y ago
Read this article and slowly re-implement the accompanying code: https://karpathy.github.io/2015/05/21/rnn-effectiveness/ It's difficult and tedious but also very useful to implement back-propagatio
936.
by bjourne 2y ago
The problem with the physicist's definition is that the larger the N the less the geometrical interpretation makes sense. For 1, 2, and even 3-dimensional tensors there is some connection to geometry, but eventually it loses all meanin
937.
by bjourne 2y ago
Well, I can write a definition that is both easier to understand and shorter than yours: A tensor is a multi-dimensional array. :)
938.
by bjourne 2y ago
Yes, they work. Cold calling and similar methods are all about the the numbers so even if you are immune to them, enough people aren't. Pesticides isn't everyone's cup of tea, perhaps you'd be more swayed by Doctors with
939.
by bjourne 2y ago
One of the major issues with LMSes is grading and cheating. You have one poor teacher's assistant that has to grade 1000 student submissions by the end of the week. How do you make this possible without compromising the integrity of th
940.
by bjourne 2y ago
I did that a few years go. It was a nice visualization up until restoring division. With more components than that the layout just becomes to cluttered to be meaningful. And it is very difficult to encode all heuristics one uses when drawin
941.
by bjourne 2y ago
The amount of information in a graph that big is on the order of 10^21. You can't meaningfully "visualize" it.
942.
PyTorch VGG implementation from scratch [video](youtube.com)
1 points by bjourne 2y ago|0 comments
943.
by bjourne 2y ago
The most important part of the article is buried in the PS:es: > Commenter Ryan points out that you could also explain this data pattern as a result of sloppy post-processing, if votes were counted correctly, then reported to the nearest
944.
by bjourne 2y ago
https://x.com/weegeedutchie/status/1817903230946337121 No one cared to listen to the Spanish-language broadcast where it is explained that the 4.6% is for the other candidates combined . The stats indeed sum up t
945.
by bjourne 2y ago
I think HN ate some characters because that code doesn't look valid. But yeah, that's how you do it. In my opinion it is not pretty (e.g., what if you have some mutable context?). I also don't see how OCaml could turn the con
946.
by bjourne 2y ago
Works if you are building one list, but what if you are building multiple? What's suggested on the OCaml site and what's taught in most of academia is to use a recursive function with accumulator arguments that are reversed before
947.
by bjourne 2y ago
Maybe. "Overall, this only shows us how far on average we go from the original distribution, but the process can only ’terminate’ if the estimated variance at a certain generation becomes small enough, i.e. we effectively turn into a d
948.
by bjourne 2y ago
Fair point about the mean, but I don't see how the random walk causes the standard deviation to shrink towards zero.
949.
by bjourne 2y ago
The article contains no proof of theorem 3.1 and finding counterexamples seems trivial. Adult male weight can be modeled by N(85, 20). You can recursively "train" the model on data it generates without having it collapse. It will
950.
by bjourne 2y ago
Do not forget all amazing discoveries made by cycling's so called cheaters. Where would the Furies, Joshuas and, dare I say it, Usyks be today if it weren't for Armstrong's pioneering work on blood doping in the 90's?
951.
by bjourne 2y ago
I wonder what people 80 years into the future will write about the ongoing destruction of Gaza.
952.
by bjourne 2y ago
What is a workflow in this context?
953.
by bjourne 2y ago
Information content is not proportional to word count.
954.
by bjourne 2y ago
You may find this site interesting: https://folkrnn.org/tune/119105 Thing is tabs and other kinds of notation only capture a small part of the composition, often just the melody. So to generate "real" polypho
955.
by bjourne 2y ago
Apache iText, for example.
956.
by bjourne 2y ago
There are of course way more efficient methods for generating templated pdfs than using a typesetter.
957.
by bjourne 2y ago
But this has nothing to do with encoding and decoding.
958.
by bjourne 2y ago
Go by alphabetical order.
959.
What Is Competitive Learning?(datacamp.com)
1 points by bjourne 2y ago|0 comments
960.
by bjourne 2y ago
What you write is eerily similar to one of the pain points of Haskell. You can write a compiler that is purely functional. But then you want logging so you must put wrap it with the IO monad. And then also every function that calls the comp

More ›