#dqrng — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #dqrng, aggregated by home.social.
-
Is 42 a "good" random seed? I know I have used it a lot in various stackoverflow posts. But this can indeed have a negative effect on people and *especially* LLMs using the example code as Claus Wilke is arguing here
https://blog.genesmindsmachines.com/p/if-your-random-seed-is-42-i-will
and here
https://blog.genesmindsmachines.com/p/random-seeds-and-brown-m-and-msI like the idea to create random random seeds and inform the user about it to enable reproducability. I might incorporate that into {dqrng}.
-
A new version of my #Rstats 📦 #dqrng has been published on #cran
Main reason was an UBSAN error from using the "modular multiplicative inverse" oy 69069 to undo R's seed scrambling. I still think it is fascinating that this unscrambling is possible at all.
I also had to disable PCG64 for MacOS on PowerPC since some parts of the included #PCG code that are now used do not work on that architecture. Once the #Debian package is updated I might find further architectures that are affected by this.
-
Version 0.4.0 of my #Rstats package #dqrng has been published on #CRAN today. At the moment we are still working through some breakage with reverse dependencies, but that should clear up soon. What I like about this release is that different streams of work are coming together quite nicely. More details in the blog post at
-
New blog post on the upcomming changes to my #RStats package #dqrng.
https://stubner.me/post/2024-04-23-upcoming-changes-dqrng/The next version of {dqrng} is going to include many changes, so it might make sense to start blogging about them before the actual release. This release will be version 0.4.0 since there are several braking changes. It also sees several new contributors. At the moment, I am in contact with various authors of packages that depend on {dqrng} and who’s tests are failing due to the breaking changes. They will have to upload new versions of their packages before I can submit {dqrng}.