krackers

26 ms·

691.
by krackers 1y ago
I'm confused as well, I thought gpt-image could already do most of these things, but I guess the key difference is that gpt-image is not good for single point edits. In terms of "wow" factor it doesn't feel as big as gpt
692.
by krackers 1y ago
https://www.sdcard.org/press/thoughtleadership/the-sd-memory... has more details, and a comparison to disk utility format. There is also this open source implementation https://github.com/profi200&
693.
by krackers 1y ago
Matformer seems a better approach to this type of scaling though
694.
by krackers 1y ago
This article seemed really basic, no insight other than "it learns the high dimensional manifold on which cat images lie, thus separating cats from non-cats" (not that simple explanations are bad, but Quanta articles seem to be ge
695.
by krackers 1y ago
I think this 2D version shows the issue clearly https://mathworld.wolfram.com/DiskPointPicking.html
696.
by krackers 1y ago
It only works for trains because the image of train at t+1 is basically image of train at time t shifted over by a few pixels, right? It doesn't seem like this would work to capture a picture of a human, since humans don't just ri
697.
by krackers 1y ago
Not the same longevity characteristics, I thought newer SSDs are TLC compared to SLC of old SSDs
698.
by krackers 1y ago
I wouldn't trust the adapters. Maybe in theory it's possible to do it right, bridge PD negotiation to a 1-wire protocol negotiation but you've got to trust that they handled both the usb-c part correctly and the 1-wire part c
699.
by krackers 1y ago
The actually interesting part seems to be adversarial images that appear different when downscaled, exploiting the resulting aliasing. Note that this is for traditional downsampling, no AI here.
700.
by krackers 1y ago
I think there was some quote from Karpathy who said that RLHF isn't actually "true" RL. As an armchair person, even after trying to understand it RLHF always seemed so roundabout. You don't have some open ended environme
701.
by krackers 1y ago
Yup, RLVR as implemented by Deepseek et al. use only outcome supervision instead of process supervision. There have been attempts to do process supervision though.
702.
by krackers 1y ago
Ever since the UI redesign they've lost the plot a bit.
703.
by krackers 1y ago
i mean you can use semaphores to implement condition variables, but it's extremely tricky to the point where you've got to be Mike Burrows in order to do it right [1] (in fact, that's how it's done in the nsync library,
704.
by krackers 1y ago
I think it's the same underlying principle of learning the "joint distribution of things humans have said". Whether done autoregressively via LLMs or via diffusion models, you still end up learning this distribution. The ins
705.
by krackers 1y ago
>Dave's toothpaste has both NHAP and flouride And has an RDA of 101. Why on earth would a toothpaste meant for sensitivity have this high of an RDA? (Non-whitening sensodyne is around the 40s)
706.
by krackers 1y ago
Maybe another analogy (or at least the way I intuitively understood it) was that humans sometimes skip over tokens we know to be fluff/filler. Without a sink, models have no way to "skip" over a token, that token _will_ atten
707.
by krackers 1y ago
This should answer https://news.ycombinator.com/item?id=44149238
708.
by krackers 1y ago
I'll never get over the fact that the grad student didn't even bother to use gpt-4, so this was using gpt 3.5 or something.
709.
by krackers 1y ago
Not your weights => not your waifu
710.
by krackers 1y ago
>gently shrinking "Gentle" or not, is this actually safe for long-term use? Your pupil dilates in indoor light for a reason, I'm not sure if it's a good idea to fight that. You'd be better off just increasing amb
711.
by krackers 1y ago
simply disabling JS seems to do it
712.
by krackers 1y ago
Alibaba Intelligence.
713.
by krackers 1y ago
> not generally used for general purpose search Possibly because up until now the performance of semantic based search wasn't worth the complexity tradeoff. I mean NLP was a hard problem, and we'd spent decades fine-tuning trad
714.
by krackers 1y ago
> application can prevent sleep _when you close the lid_ Ordinarily it can't, it's not possible to set a IOPMAssertion that prevents sleep on lid close. That's probably one of the reasons why the sleep experience on mac is
715.
by krackers 1y ago
Hm I'm still not sure I understand what you mean by "sets up and tears down the entire main loop". Even in normal cocoa applications every keystroke or input event causes an event to be dispatched from WindowServer to the app
716.
by krackers 1y ago
I'm not sure if I understood the issue right, but if the issue is with the runloop, I think you can technically reimplement it yourself with nextEventMatchingMask. But their design just seems broken, if they're re-initializing the
717.
by krackers 1y ago
The paper showing that dropping positional encoding entirely is feasible is https://arxiv.org/pdf/2305.19466 . But I was misremembering as to its long context performance, Llama 4 does use NoPE but it's still inte
718.
by krackers 1y ago
I've read that "No Position Embedding" seems to be better for long-term context anyway, so it's probably not something essential to explain.
719.
by krackers 1y ago
Some people will adapt, the same way people adapted to flamebait. But even today new marks still end up taking the bait. The more annoying issue to me is that the kind of sycophantic writing style seems to end up spreading to all LLMs. May
720.
by krackers 1y ago
Using science fiction as a basis for philosophy isn't wise, especially TNG which has a very obvious flavor of "optimistic human exceptionalism" (contrast with DS9, where I think Eddington even makes this point).

More ›