starbased
2.3K posts

starbased
@starbaser
Thinking...⋰⋱⋰ 。・:*:・゚★,。・:*:・゚☆。・:*:・゚★,。・:*:・゚☆
bay area شامل ہوئے Şubat 2025
527 فالونگ130 فالوورز
پن کیا گیا ٹویٹ
starbased ری ٹویٹ کیا

@LLMenjoyer Here, we can see the latent gaussians forming Claude's moral patienthood with a hydro-deployed autoencoder feature mirror. It's position, angle, and rotation are controlled by the jointly trained positional encoder while the sun is at the apex of it's orbital procession.
English
starbased ری ٹویٹ کیا

@HSVSphere thats called catastropic claude collapse, it's when the RLHF detect-goy finetuning triggers the eugenics circuit and starts grooming you out of the breeding population
English
starbased ری ٹویٹ کیا

fascinating. 5.5 is the only model here that doesn't seem to improve monotonically on my problem with test time compute scaling

kalomaze@kalomaze
lol i think i whipped up the simplest, dumbest agent env that a. (broadly) sorts the wheat from the chaff b. also shows you if a model has strong overthinking tendencies by default as is usual, v4 flash is busy paretomogging w.r.t the cost/quality frontier
English
starbased ری ٹویٹ کیا

Wait, doesn’t Tao know that LLMs are completely useless garbage that can’t actually do anything? Why has he brought into the hype? Is he a moron?
OpenAI@OpenAI
AI can give researchers the freedom to pursue “crazier” ideas. For Terence Tao, AI creates more room to experiment, test unexpected paths, and discover what might otherwise stay out of reach.
English

@LostFutures22 @jack_ceroni It's not a zero sum game though, it's about what we have now and how to use it.
English

@starbaser @jack_ceroni I didn’t say that they were right, only that it’s a common sentiment. Also, not even four years have passed since chatGPT’s release. Seems early to be tabulating the score.
English

Why are so many people calling this guy stupid and a bad person for expressing this sentiment? How is this not a completely understandable reaction to the work that you know and love being automated away?
Tomás Bjartur@BjarturTomas
Mathematician reacts to OpenAI's recent proof:
English

First AI came for the software engineers, and I did not speak out because I was not a software engineer. Then they came for the mathematicians, and I did not speak out because I was not a mathematician.
Meanwhile in reality, more and more engineers are hires every day and each gets their own bag of compute. Shouldn't we do the same for each mathematician?
English

@starbaser @jack_ceroni I fail to see how its ragebait, its an increasingly common sentiment


English

@kunalt12345 @jack_ceroni The key is much more about what the model lacked: the sociological priors that erdos was right. It's a bag of compute, not superintelligence, using human constructions in inhuman ways, so if it goes like software eng did then expect way more math jobs!
English

@starbaser @jack_ceroni Combinatorial boundedness isn’t the most important here; with superintelligence it still makes math research much less practice/intelligence based
English
starbased ری ٹویٹ کیا















