Last Fraction Zero

@NoorInvesting @elonmusk Being the only currently viable commercial provider does have its perks, doesn't it?

@elonmusk Still doesn't justify the target market cap of $2 Trillion with annual revenue of $20 billiom whilst $META has 1.5T with revenue of $135 Billion

This trend will continue for many years

@dukesofbastard @Gral_Julian_25 @GreenTextRepost There's a Devo interview where they talked about a faulty synth that produced a very specific sound they liked. When it finally stopped working altogether they had a difficult time finding someone who could understand they wanted it repaired but not actually fixed.

@Gral_Julian_25 @GreenTextRepost I had something similar done. I had a guitar that I set up, it was rough looking, etc. A friend took it to a guy in town and the guy in town fucked my intonation, put on the wrong gauge of string, and screwed up the wiring. He did it the correct way, but my guitar wasnt for that

@I_am_Superbro @BarronTNews_ The pinnacle of Machiavellian geopolitical fuckery: Placing civilian infrastructure directly adjacent to a military base, less than 300m from the closest military structure, using it as a school, and specifically a school for girls. You are the proof that it works as intended.

@BarronTNews_ I don't give a f'ck. Trump assassinated 170 Iranian daughters when he dropped a Tomahawk on their school. He reaps what he sows.

Lets pray together🙏 Heavenly Father, cover Ivanka with Your peace that surpasses all understanding. Guard her heart, her mind, her home, and her family. Send Your angels before her, behind her, and beside her. Let her walk in safety, strength, and courage, knowing that You are her refuge and fortress. We thank You that she is safe. We thank You that darkness did not prevail. And we ask You to continue protecting President Trump as he stands in Washington carrying the burden of this nation during such a serious time. The Lord is our light and our salvation. Whom shall we fear? God protect Ivanka. God protect President Trump. God protect the Trump family. God protect America.

@sleepy_devo @AlbaLink @ArtemisFowl_003 @Sargon_of_Akkad Shit, the framework predicts direct correlation between income and birth rate but the actual data is inverse. I know what to do! When someone brings up income's inverse relation to birth rate I'll narrow scope to the middle class and ignore the peak at low and valley at high.

@AlbaLink @ArtemisFowl_003 @Sargon_of_Akkad it actually does. a middle class family cannot afford to have a dozen kids, and they are smart enough to know that in 18 years time, college funds will be due. a lower class family cannot afford to have a dozen kids, but they don't think on a time scale of decades.

the reason that women aren't having kids is a lot more complicated than you're making it out to be, @Sargon_of_Akkad - lack of financial stability - lack of political stability - a growing number of men who are frankly, not breeding material it's not as simple as "libs bad"

@emzanotti IHOP HATES HIM! One simple trick to get free pancakes for a week!

@emzanotti Yes, bunch of them actually argue frontload cost goes 100% into the first portion used. They will absolutely have a public melty if you mention their math means using a second portion tomorrow = free food. At this point I mock the one who won't divide cost: "I get free pancakes"

Oh my god you don’t just use the items once, and you don’t have to buy $12 mayonnaise. I started out defending these people but, you know what? They deserve whatever it is that’s happening to them.

@ILYA_babay @nick_kapur EBNF looks like it can be used for templating an ontological map; enforced section headers and formal structuring of lines into prose (free generation) and datatype (verified facts) segments. I would hope leaving prose segments free wouldn't break phrasing but that needs testing.

@ILYA_babay @nick_kapur Honestly, not sure. I read about it when trying to theorycraft what your pipeline might be. I assumed ontological mapping of a clinical notation framework (Eg: SOAP) + datatype libraries (symptoms, diagnoses, rx, etc) are used when building and populating relational tables. 1/2

An auditor for the Ontario, Canada government found that AI agents tasked with turning doctor/patient conversations into structured notes routinely hallucinated false treatments, replaced drug names with entirely different drugs, and missed crucial information

@makacre @whoooooowhooo @MatrixMysteries Where the hell did you get 2hr ER wait times? I sliced my hand open in the early 90s and spent 6 hours in the waiting room while listening to a half dozen lectures from my parents about being careful while holding a glass. Every other time has been 8+ hours.

@whoooooowhooo @MatrixMysteries Nice logic. How about we go back to 5 years ago when you could get free health care with a 2hr wait at the ER. Not an 18hr wait

A Canadian woman with a swollen appendix was told to go to the ER. She films the wait-time board. 15+ HOURS to see a doctor. And she says she had already been waiting 3 hours. This is “free” universal healthcare in Canada.

@ILYA_babay @nick_kapur Ahh, I get it now. This is, roughly, the pipeline you're describing?

@LastFraction0 @nick_kapur Fair — NL fluency unavoidably uses weight-tokens for grammar and coherence. The discipline lives on the claim-carrying tokens: every factual assertion has to trace to a retrieved chunk. Weights handle the connective tissue; entailment check catches the drift.

@MelbaofNM I see two of the sub-loans in the loan package paid off, or nearly paid off and a green bar below payments indicating that 34% of the value of the principle + total expected lifetime interest (if she keeps paying at that low rate) has been paid off already. What's confusing?

@ILYA_babay @nick_kapur What you're hoping for is an outcome where the report is actually a template where wildcards are filled in only with tokens from the documents being consolidated, but, the template creation itself is part of the synthesis and that risks inclusion of unwanted parameters.

@ILYA_babay @nick_kapur Except it was observed in training data. It's ability to output in natural language necessarily requires it to rely on its parameter weights. It can't output that way without synthesis drawing tokens from the pool of training weights in addition to the documents being compiled.

@triiadxx One the left: things largely created by people who learned via apprenticeship-style education and were referred to as apprentices, journeymen, and masters. On the right: things largely created by people with degrees.

@ILYA_babay @nick_kapur This isn't merely data entry into fixed fields where it can pull relevant strings and record them into those relevant fields. This is whole cloth natural language synthesis. It has to generate from training data if you want output that isn't just a list of relevant strings.

@nick_kapur The fix is architectural: generate only from what was actually recorded, never from training data. AI that fills clinical gaps from learned patterns will always swap drugs and invent treatments. Retrieve-first, synthesize-from-retrieved is the only durable defense.

@RepMGP If you decided the measure of success is determined by flavor, not nutritional content, then you have lost. In terms of flavor every apple variety you know will lose to a bag of processed carbs, fats, and sugars because those products are engineered to taste better than produce.

Given that Red Delicious are the worst apples, have you ever wondered why they’re in every school cafeteria? It’s because there’s a list of "approved" apple varieties allowed in schools, and it hasn’t changed since Kmart sold firearms.

@bugcel_ Huh, that actually sounds relatively decent for a boarding house arrangement. All the ones I lived in where run by landlords that didn't care about anything except rent being paid on time.

they have a once a month maid come in, i’ve only seen them like 2 times i think. Usually it’s the tenants responsibility, there’s a few rules that i blurred out in the video for privacy but basically quiet hours are 10 on weekdays and 11 on weekends, don’t leave personal belongings out, clean up immediately after using something. Pretty basic stuff. They have some stuff on the “lease” too, idk it’s called a membership agreement. No bitcoin mining lol

@fabiolean @EconTraina @DellAnnaLuca It was facetious. Meant to highlight that identifying hallucinatory elements and compiling incontrovertible evidence involves several people putting in unpaid hours, and, costly access to paywalled databases because AI will generate fictive citations attributed to real papers.

@LastFraction0 @EconTraina @DellAnnaLuca Nah. They should have to take the ban. Having a wealthy sponsor for your research shouldn’t exempt you from basic shit like “knowing what’s in your own paper”

The reactions of many researchers on finally being held responsible for having read the very paper they submitted are... something.

@dpf11100 @petruliun @BladeoftheS You're so close to getting it. Demand that the government does more for everyone using only the tax revenue they are already collect. Otherwise you're inviting them to further increase taxes to pay for the same promises they previously increased taxes to pay for and then broke.

@petruliun @BladeoftheS The solution to your problem is demanding that the government does more for everyone, instead of complaining about other people getting support. Because one leads to more support for everyone, the other leads to no support for anyone.