Big news from JuliaPackages.com. We’ve given the site a major refresh - the first in two years!
The headliner? I’ve collaborated with OpenAI’s ChatGPT to categorize a whopping 1,000+ packages. No more reliance on svaksha’s decibans project; with AI-powered categorization, your perfect package is just a click away.
Drop by, check it out, and share your thoughts. I’m all ears.
// written for/with/by ChatGPT
Wow, that turned out really good. Thanks for sharing. The only thing that immediately pops out to me is that it’s odd that ModelingToolkit got classified under Programming Paradigms rather than Mathematics. Other than that it looks rather solid.
Thanks for the effort! I agree that the categorization looks fantastic and highly accurate! I want to learn how the trending page is calculated!
This looks really nice!
I would also be interested in learning more about the categorization, since there seems to be a category Optimization (on its own, not a sub of Mathematics), where Manopt.jl is in, while it is build upon ManifoldsBase.jl, which is Mathematics (sub geometry) - and quite fitting. So if multiple categories would be possible, that would be neat, since Manopt is half Optimization, half Math/Geometry.
But overall this looks really useful!
Looks like somehow JuMP got categorized as a “server” instead of optimization: JuMP · Julia Packages
Nice! I imagine getting the initial categorization done without automation would not scale.
Is there a description of the categories somewhere? I couldn’t find one. https://juliapackages.com/p/makie is filed under “non-graphical plotting”, and I’m not sure what that is supposed to mean?
Curiously, in contrast, Plots · Julia Packages is under Infographics, and I would intuitively have expected the same category for both.
Will look all this over. Think we can solve a decent amount of these issues in the next version
How often is this updated? A package of mine that’s been registered for about 3 weeks now (FieldFlags.jl) doesn’t seem to be indexed.
Oh hey, this actually renders my README’s (mostly) correctly, see DataToolkit - JuliaHub vs. DataToolkit · Julia Packages.
I do notice that the image URL isn’t quite right though, it has
https://github.com/tecosaur/DataToolkit.jl/raw/master/docs/src/assets/logotype.svg, when it should have
Did you run the AI thing once, or do you repeat it to refresh data? If so, it looks like you’ll need to run something to do manual fix ups of the mistakes. Hopefully that list is not too big. Just looking at Mathematics I find, in addition to those mentioned above, the following are not correct, or are doubtful
It looks like only a small fraction are miscategorized.