-
got distracted and generated some fictional names for discontinued perfumesPermalink Look on archive.org
♻️ 4 Retweets ❤️ 31 Favorites Mood -2 🙁
-
using: - the list of perfumes that scentmatchers.com/ claims to be able to make dupes of - some regex in Sublime Text to clean the source - and textgenrnn running in google colab minimaxir.com/2018/05/text-neural-networks/Look on archive.org
-
it's pretty neat that you can do all this in about half an hour! much easier than getting char-rnn installed & configured locally.Permalink Look on archive.org
-
fun fact: if you don't generate with temperature set to 1.0 (translation: tell the computer to pick less-likely options), then it thinks every perfume is by AvonPermalink Look on archive.org
-
(now i am looking at the generated list and imagining a person who would wear each perfume)Permalink Look on archive.org
-
choose a fictional person to date based upon the fictional perfume they wear:Permalink Look on archive.org