paliforever

ARSENAL - 2026 CHAMPIONS, 76ERS

[paliforever @paliwali33]

@_trundler @natolambert that's one of them yeah

[paliforever @paliwali33]

@LeGarconCalme @benhatesnufc gtfoh

[paliforever @paliwali33]

@HandofArsenal @Knifesedge11 SPEAK ON SKELLY PLEASEEEEE

[paliforever @paliwali33]

@cuoium I don't think that Bastoni is better than Gabriel tbh, maybe someone like Pacho?

[paliforever @paliwali33]

@Altimor This is a bizarre conclusion to reach. Jargon is a way to hide unintelligible behavior / lack of understanding. It's a big mistake to take the model's word as law or view them as superior intelligence. You can also ask them to explain more simply and they frequently will

[paliforever @paliwali33]

@trashbaby40k @tautologer it seems that projects like AI2027 just kind of take various obstacles (continual learning, infinite context, etc) like they will inevitably be solved... i think they probably will be but it could take much longer than expected + hardware codesign might change too

[paliforever @paliwali33]

@IronEMK @Footballtweet PSG play in a joke in a league, consistently enter EOS fresher than everyone else

[Billy Carpenter @billycarpy]

I'll say this: when watching Vini, it's easy to get distracted by the dribbling and the goals and the speed and the passing and the movement. But I was watching some more over the last 24 hours, and actually, he sucks.

[paliforever @paliwali33]

@francoisfleuret why do you think any alien civilization wouldn't have their own much more advanced agents?

[paliforever @paliwali33]

@BiblioPhilip @CaptionQuirk I don't understand how this is even a potential reading of that scene, it's actually insane

[paliforever @paliwali33]

@SourceCodeplz @SullyOmarr difference between efficient & unreadable, this is clearly closer to the latter. have heard some say its a consequence of too much RLVR which seems v plausible.

[paliforever @paliwali33]

@cstrlke32 @_Ventfull_ he's the same guy, rohan from tdk confirmed

[paliforever @paliwali33]

@KrabbyKoder @frogs4girls they're autoregressive sure but calling them markovian kind of beats the point. its like calling them an n-gram model w/ n = every previous token.
LLMs can and frequently do generalize beyond their training dist. that much is undeniable (the extent is arguable)

[Nick @BigBicNick]

Why are conservatives randomly talking about Covid 7 years later and not how awesome the economy is and how great their war with Iran is going

[paliforever @paliwali33]

@karthikadhaigal good news = tier 1

[paliforever @paliwali33]

@KrabbyKoder @frogs4girls LLMs are not Markovian. The response is a conditional probability distribution but lots of evidence the models can generalize beyond training dist - unclear what the implication is here tbh

[paliforever @paliwali33]

@Afcwill19 @afcjxmes maybe the club but bruno is 28.. he surely doesn't want to be in a relegation battle w/ a new coach, no isak, no gordon, etc. fans won't like it but he surely can see the writing on the wall

[paliforever @paliwali33]

@flawedaxioms incredible bait well done

[Mo @mo_gadar1]

Now that the seasons over let me say this is one fucking hard edit