Vecr

33 ms·

511.
by Vecr 2y ago
Is that an ESD hazard? Do you have a special antistatic hoodie?
512.
by Vecr 2y ago
That's not really how probabilities work. There's so many things happening in the world that you expect some things that are individually unlikely to happen to happen almost all the time. You just don't know in advance what t
513.
by Vecr 2y ago
No it's not, I agree, except I think they're worse off than housecats. Not that cats have it the best: https://gwern.net/review/cat "Cat Psychology & Domestication: Are We Good Owners?"
514.
by Vecr 2y ago
It's quite bad? It's the only one I've ever used.
515.
by Vecr 2y ago
There's no abort system at that point. They probably could have saved Columbia though.
516.
by Vecr 2y ago
I don't really understand, "paperclips" is a stand-in for anything that would make the universe have near-zero value (when tiled/converted to this substance/pattern) when evaluated as a hypothetical in a public poll
517.
by Vecr 2y ago
Is that actually better than just cutting power to all Internet infrastructure?
518.
by Vecr 2y ago
Only the LED light bulb. And then only kind of, they use approximations.
519.
by Vecr 2y ago
That's a big problem though, mechanistic interpretability/explainability is really important for the whole "not wrecking everything" part of AI. You could also just hope AI stalls out or goes back to more hand-tuned Baye
520.
by Vecr 2y ago
Eternal Golden Braid, that is. Quite a bit of it's outdated at this point, but the explanation of uncomputability is probably still the best.
521.
by Vecr 2y ago
Fractional reserve backing is old school, zero reserve banking is where it's at.
522.
by Vecr 2y ago
Well, most terrorists are not very smart, at least on an absolute level. I think even near-future models would be able to help them on a basic operations/not screwing up level. As I said, a problem is the reconstruction of information
523.
by Vecr 2y ago
Yeah, with this kind of stuff it's just a reputation thing. In future systems it's a problem though, if you accidentally train them on on the wrong thing or the system is smart enough to reconstruct the missing data, the model can
524.
by Vecr 2y ago
If you download weights and keep them around this kind of thing works forever. Well, until the model starts showing its age due to the cutoff date.
525.
by Vecr 2y ago
Logo looks kind of Star Trek. So does the ?cat? creature. The early animated show I guess? Is Stars Reach a unique enough name though? I'm getting results with and without the apostrophe.
526.
by Vecr 2y ago
You should see some of Schmidhuber's stuff, he uses scientific journals as his blogging host.
527.
by Vecr 2y ago
Even Numbers 31?
528.
by Vecr 2y ago
There's a lot of extrapolation going on already. It's like if you speculated an evil demon as running a VR simulation in the year 1900, more powerful than current computers. No knowledge of transistors needed. On the level of math
529.
by Vecr 2y ago
Those aren't children's cartoons. I mean, maybe they are, but think more Zootopia squished into weird Dilbert.
530.
by Vecr 2y ago
Numbers too?
531.
by Vecr 2y ago
How hard would it be to keep the thing drained at all times?
532.
by Vecr 2y ago
I know posting "I agree" is not generally welcomed on here, but ChaCha8 is really underappreciated as a MCMC/general simulation random number generator. It's fast, it's pretty easy on cache, and it won't bias y
533.
by Vecr 2y ago
The model weights themselves can't do that, it's just not how single-pass token-by-token generation works. Even if you got rid of tokens and went byte-by-byte there's no improvement. To do what you are saying you'd have
534.
by Vecr 2y ago
As far as I know there's nothing in the middle.
535.
by Vecr 2y ago
That's something a hosted product could do, but if you have access to the weights you just don't run the additional filtering code. Also, the filtering code would be expensive to run, if you want it to be good quality. The result
536.
by Vecr 2y ago
It's below the Armstrong limit, you don't need pressurization. At this altitude occasional oxygen is probably fine. Pressurization is mostly a comfort thing until you get pretty high up.
537.
by Vecr 2y ago
The atmosphere's a pretty good radiation shield. You need air filtration so you don't come in direct contact with radioactive particles, but that's pretty much it. Flying gets you away from ground burst radiation by default a
538.
by Vecr 2y ago
The original proposed mechanism was probably wrong, these drugs probably make you change your diet. Or at least eat less, obviously you have to be concerned with nutrition.
539.
by Vecr 2y ago
Where do you find princesses anymore? Sweden?
540.
by Vecr 2y ago
Not really, you erase and then inpaint, or replace the offending area with multicolor noise, put a bit of multicolor noise over the whole thing, then do an image-to-image run on that.

More ›