Now that it is getting interesting with the roman numerals :P
bitmap needs a guess case button for his commit titles
ApeKattQuest joined the channel
ApeKattQuest has quit
ApeKattQuest joined the channel
ApeKattQuest
thanks reo :D
also sorry for my flakey connection eveyone, internet provider decided to hav an issue on 17th may, i guess
they *did* message me about having it fixed, but i was afk during this ,and only developed issues after the fact
well well
zas
yvanzo: are python -m sir amqp_watch processes properly killed on template changes in sir-solr9-prod? there were 1k+ of those on rakim before I restarted the container
lucifer: ^^
reosarevok
kellnerd: "Guessed language of the titles: Estonian (42% confidence), English (16% confidence), Turkish (15% confidence), rn (11% confidence)"
it's English :D
Cool that Estonia is first in something, but wat :D
kellnerd
I know that the accuracy for few/short titles is bad, but the library at least has confidency values :P
reosarevok
Ah, ok, yeah it's a one-track single
"RIHANNA FREESTYLE - Single" - for Apple imports, you should probably auto-drop "- Single"
kellnerd
Yeah, currently there is no cleanup at all, just pure data
I still to think how data cleanup best fits into the architecture, probably a provider-specific and a generic cleanup stage.
atj
reosarevok: the model isn't optimised for small amounts of text unfortunately
but it has to be > 70% confidence to be automatically selected IIRC?