coldtea

29 ms·

541.
by coldtea 4mo ago
The gravity in the paper, as far as I can see, is about the influence of microgravity (e.g. much much smaller than regular Earth gravity) influence in blood circulation, bodily fluids etc, and as such on sleep, breathing, heart function, an
542.
by coldtea 4mo ago
> A bad abstraction would at least have had one fire in one place. On the contrary: that's precisely what a bad abstraction would not offer. Instead it would spread its assumptions to different parts of the system, as every caller,
543.
by coldtea 4mo ago
If it was Sturgeon's Law it would be a constant across different eras. But different times absolutely can have worse production.
544.
by coldtea 4mo ago
Hardly iconoclastic, it's a very sensible suggestion. It would be iconoclastic if the common sense basic approach would be to start with abstraction. It's not, the common sense default is to write possibly duplicate behavior until
545.
by coldtea 4mo ago
He's making the point that the idea that "with a human you can more or less dump them everything you sit on, and let them shift it through, and they'll mostly make it out OK" is wrong. Humans absolutely need lots of trai
546.
by coldtea 4mo ago
> their mistakes aren't based on their understanding, it's basically random guesses. Not random across their whole training set. Random across related concepts bundled together in the training set. Which is not that dissimilar
547.
by coldtea 4mo ago
> I have a theory- LLMs can spit out code that gets the job done and looks ok, maybe even great, but contains small “anomalies” that compound over time. An enterprise app developed entirely with LLM-happy devs might end up virtually unma
548.
by coldtea 4mo ago
> I suspect (ignorant armchair) we have some ability to signal strength of those facts, via repetition. Without this layer of introspection Isn't that precisely what the LLM training does? It signals strength of those facts, via re
549.
by coldtea 4mo ago
> The original intelligence that created those tokens was driven by a whole universe of inputs, from hormones to starlight to gravity Still inputs, that in the end changed something about synapses and their activation. And whether doesn
550.
by coldtea 4mo ago
How is it's process (inputs, training weights, processing those weights based on an initial given state -prompt-) different than how a brain thinks? And is it a difference of architecture and scale (like an ENIAC and a Apple Silicon M5
551.
by coldtea 4mo ago
"Company execs and stuff pre-IPO declare their product is amazing and will change everything, news at 11"
552.
by coldtea 4mo ago
> Perhaps the smartest researchers in the world across multiple labs know more about this than we do? Perhaps the smartest researchers in the world across multiple labs follow the money, and don't make waves that go against them ge
553.
by coldtea 4mo ago
> We have observed consistent scalings of language model log-likelihood loss with non-embedding parameter count N, dataset size D, and optimized training computation Cmin, as encapsulated in Equations (1.5) and (1.6). Conversely, we find
554.
by coldtea 4mo ago
The same company that let their flagship product go to waste, then bet hundreds of billions into the stupid vr metaverse that went nowhere? I'ts not like they really inspire much confidence in their future predictions.
555.
by coldtea 4mo ago
Sound like the very definition of marginal returns and/or desperation, combined with throwing money at the problem...
556.
by coldtea 4mo ago
> There are journalists being hired to write Atlantic-worthy articles that exist only as LLM training data, because they're getting paid more than the Atlantic would pay them for it. So now part of the training is directly slop - j
557.
by coldtea 4mo ago
What's funnier is the implication under which the "objection" is stated, as if those researchers didn't/couldn't know it.
558.
by coldtea 4mo ago
> one generation can't "get" the entertainment of the next generation. It has always been like that. Given the quality of 2026's entertainment, looks like they had a point. And likely they had one at 2006 and 1986 an
559.
by coldtea 4mo ago
> I would absolutely draw a distinction between government enforced censorship and private platforms censoring content. I would absolutely not.
560.
by coldtea 4mo ago
> These are wild claims - why are we concluding that bigger models and more data = more hallucination? Because that's what they measured in this case.
561.
by coldtea 4mo ago
Let's start by firing the author.
562.
by coldtea 4mo ago
Not that massive. A lot of good, creative ideas have been called out and derived as nonsense or crazy. Many still are.
563.
by coldtea 4mo ago
As opposed to what? Water and tissue?
564.
by coldtea 4mo ago
"self-driving cars"
565.
by coldtea 4mo ago
> By that logic regulators should lower fines if the action wasn’t profitable. The logic isn't some rigid "make the fine based on the profit". The logic is based on the intent: make the behavior happen less. So you can ha
566.
by coldtea 4mo ago
Well, isn't that my point exactly? So the parent saying "The fine is only part of the story. They likely spent more money than the fine fighting it over 5 years as fines increase next time if you don’t stop" doesn't inva
567.
by coldtea 4mo ago
Precisely to avoid the custom NIH Polars API, and use SQL which works everywhere (yes, inconsistencies aside).
568.
by coldtea 4mo ago
> I don't wanna badmouth Java people, but how they push the idea that this thing is some sort of genuine breakthrough that took multiple PhDs years of cutting-edge research to implement, when in fact they basically copied what .NET
569.
by coldtea 4mo ago
This is common experience also in ND vs NT differences.
570.
by coldtea 4mo ago
We were better off when he wrote lexers

More ›