krackers

29 ms·

571.
by krackers 11mo ago
I thought Threads was already the next generation of zigbee.
572.
by krackers 11mo ago
One could say that it depends on your basis...
573.
by krackers 11mo ago
Yes, you're right. I typed "**" (exponentiation) but HN ate the second star since I forgot to escape.
574.
by krackers 11mo ago
I always like to think LLMs are markov models in the way that real-world computers are finite state machines. It's technically true, but not a useful abstraction at which to analyze them. Both LLMs and n-gram models satisfy the markov
575.
by krackers 11mo ago
https://www.lesswrong.com/posts/gTZ2SxesbHckJ3CkF/transforme... explains this more, see the part about mixed state presentation & belief synchronization >Another way to think about our claim is that transfo
576.
by krackers 11mo ago
Is there a summary? Every time I try to understand more about what LeCun is saying all I see are strawmans of LLMs (like claims that LLMs cannot learn a world model or that next token prediction is insufficient for long-range planning). The
577.
by krackers 11mo ago
see the broken window fallacy
578.
by krackers 11mo ago
I think it's hilariously tone deaf that travel booking and shopping are the two examples of "agentic" AI that keep popping up.
579.
by krackers 11mo ago
It's funny that most intro calculus courses will make it a point to remind you that "dy/dx" isn't a fraction, then when they get to integration & diffeqs they want you to forget that and start manipulating them
580.
by krackers 11mo ago
Companies mass crawling don't use LLMs for crawling itself, that would be too expensive.
581.
by krackers 11mo ago
https://www.lesswrong.com/posts/jGuXSZgv6qfdhMCuJ/refusal-in... provides more detailed information on the theory behind abliteration
582.
by krackers 11mo ago
MCP is something that's filled with buzzwords and seems like something created solely so that you can be "sold" something. From what I actually gathered, it's basically somehow four things rolled into one: * A communicat
583.
by krackers 11mo ago
> FFT has periodic boundary conditions FFT is simply an algorithm to efficiently compute the DFT. The fact that the article makes no mention of DFT vs fourier series vs DTFT is going to end up creating more confusion that it solves. For
584.
by krackers 11mo ago
There is the platonic representation hypothesis, which speculates that as LLMs get larger and more multimodal, they all end up learning isomorphic representations of reality. Maybe for humans something like this is true as well, since ultim
585.
by krackers 11mo ago
There's a chapter in Effective Java that deals with this, and it cites https://www.cs.tufts.edu/comp/150CBD/readings/snyder86encaps... so at least as early as 1986
586.
by krackers 11mo ago
Does this not do what you want? https://developer.apple.com/documentation/iobluetooth/ioblue...
587.
by krackers 11mo ago
Every time a model is about to be released, there are a bunch of these hype accounts that spin up. I don't know they get paid or they spring up organically to farm engagement. Last time there was such hype for a model was "strawbe
588.
by krackers 11mo ago
Someone should try it and create a new "mysterybench". Find all mystery novels written after LLM training cutoff, and see how many models unravel the mystery
589.
by krackers 11mo ago
Whatever idea yann has of JEPA and its supposed superiority compared to LLMs, he doesn't seem to have done a good job of "selling it" without resorting to strawmanning LLMs. From what little I gathered (which may be wrong), h
590.
by krackers 11mo ago
>Picking QLC over TLC allows them to maintain costs while achieving sufficient endurance for cold storage. How does that work, doesn't QLC have less write endurance?
591.
by krackers 11mo ago
There's no need; they already have a polyfill for XSLT I believe, they could ship that as part of the browser. Or compile libxslt to webassembly
592.
by krackers 11mo ago
Possibly same reason why youngest people had highest rates of myocarditis
593.
by krackers 11mo ago
> is just annotation and labeling kind of work Given that this is presumably all data used for RLHF'ing, I wonder how much of this is responsible for things like LLM "sycophancy" or the issue of "hallucinations".
594.
by krackers 11mo ago
>finance/accounting to be vulnerable Aren't there certifications and stuff? I'd think that from a regulatory and legal compliance perspective it'd be better to have Americans handling the books. Say someone turns out
595.
by krackers 11mo ago
For context, I think this is the same infamous https://en.wikipedia.org/wiki/He_Jiankui https://knowyourmeme.com/memes/people/jiankui-he
596.
by krackers 11mo ago
The gpt mentioned in the article seems to be the infamous sycophantic 4o. They've probably dialed it down since then, but the timelines mentioned do seem to fit with others' anecdotal observations that the level of sycophancy star
597.
by krackers 11mo ago
>They can't be emergent generation, surely It is. It's what you get when you RLHF for catchy, agreeable, enthusiastic responses. The content doesn't matter, it's the "style" that becomes applied like a coat
598.
by krackers 11mo ago
I had no idea you (ceo of hmtx) were a professor. Do your students know that you live a double life writing banger tweets?
599.
by krackers 11mo ago
This needs to be framed somewhere, very lucid explanation
600.
by krackers 11mo ago
> then I do so and start quietly applying In this job market? And where pretty much every company seems to be following the top-down push for AI-driven "velocity"?

More ›