So, I had cause to do an emergency hot wash yesterday and in the ensuing rush I accidentally included my new #Casio #GShock #GBD300 into the load…

An hour and a half later in a 90 degree wash, did it survive? 🫣

Yes of course it did! (Though it recorded 1124 additional steps 😅)

I wouldn’t recommend *deliberately* doing this, and I’m not sure if the B2100 with its mechanical parts would have fared as well, but I’m mightily impressed!

@carlesp Rosalia com Estopa i d’altres son els principals culpables del descens en l’ús del català. Ells que podrien fer que el català fos conegut mundialment, l’amaguen.

@rockmastermike @KevBot Yeah! I have a bunch of other search terms I use when I find several clusters and want to narrow it down to the "cool" area, but in this case there was pretty much only one cluster so I stopped there.

@gabrielesvelto even this amount looks like an overkill. It's enough of you make a cross or something like that

@david_chisnall Most of the examples in the other replies are words borrowed from other languages. Whether the accent comes along for the ride is a matter of style. Many borrowed words lose their accents in English, or at least American English.

Your example of “naïve” uses a diaeresis (ok, a diacritic mark, not an accent, as another respondent pointed out) to distinguish two separate vowels from a vowel digraph. Other examples are coöperate, coördinate, reënact, reënlist, etc.
1/

@liaizon @gwil @adz @cwebber miss you all so much! 🥹sending hugs!

@iftas@mastodon.iftas.org that's true too. On the receiving end it's hard to judge whether allowlist federation is the right move.

Could be something in the middle is the solution. Filter/queue activities from unknown instances until an arbitrary number of activities have been cleared by the local admin.

@johnhowesauthor Short story collections don't sell as well as novels. And of the two novels you propose, one is SF (outsold by fantasy by a wide margin) and the other could plausibly be pitched either as Romantasy (hot *right now*) or something more innovative but harder to market.

@sundogplanets FFS. We don't deserve to survive. We are such idiots ☹️

@gabrielesvelto 😱

Una bona reflexió de Toni Soler sobre el català

@phillycodehound oh my

@neil Don't get shot up the kilt!

Yep! Costco is one of the few stores I search out while traveling overseas as they employees are super helpful & happy.

👉🏾 #Costco makes big investment to keep members coming back https://www.thestreet.com/retail/costco-invests-employee-retention-to-keep-members-coming-back

Jeg rives mellom å se på dette som sportsvasking av et fascistisk regime, på linje med å delta på OL i 1936, og å synes det er så flott og bra med alt folkelivet og gleden og det sosiale fellesskapet folk får ut av det.

#NorskTut

Sind hier Baumkenner*innen, die mir sagen können, ob das bei dieser bis Donnerstag schönen Robinie wirklich nach Baumfäule und drohendem Umfallen aussieht, wie vom Sägetrupp behauptet?

@davidgerard And don't you still need to boil the ocean to produce "local" models anyway?

@david_chisnall

(settling into my comfy IANAL armchair...)

Maybe, but it complicates the heck out of any trial. Proving workplace discrimination is already outrageously difficult. Now you're facing the additional challenges that the defendant has no discoverable documents describing their arguably discriminatory decision or biased decision-making process. Their contracted ATS AI probably has no useful and discoverable logs, either.

So what are you left with? Maybe you could convince a judge to make the model and its training data available through discovery, but we're talking trade secrets and an unreasonable amount of data. And at best you'd end up with some situation where an expert witness says the model is biased in general when examined in a lab but you can't prove that this specific decision negatively impacting the plaintiff was due to bias. And even getting this far would be prohibitively expensive for any real world plaintiff. And the LLMs themselves are often nondeterministic even with the exact same input, which is often *by design* (i.e., temperature) for aesthetic reasons.

Am I missing something here?

CRACK032.BMF