jampekka

40 ms·

901.
by jampekka 2y ago
If it's so important to us, perhaps we should support it then? The discussion around IA nowadays seems a lot like random users ranting at open source maintainers in Github issues.
902.
by jampekka 2y ago
Maybe not having to learn to write "properly" means more bandwidth for more general thinking? At least not having to care about arithmetic leaves more time to care about mathematics.
903.
by jampekka 2y ago
Rust unwraps many such problems.
904.
by jampekka 2y ago
That's the only benchmark on the page that uses such wrapper. Buffer.allocUnsafe just allocates the memory without zero-initializing it, just like e.g. malloc does. Probably usually not worth it, but in a benchmark it's comparable
905.
by jampekka 2y ago
In benchmarks JS is usually well within magnitude (i.e. under 10x slower). E.g. C++ vs Node.js here: https://benchmarksgame-team.pages.debian.net/benchmarksgame/... Couldn't find C vs JS easily with the new benchm
906.
by jampekka 2y ago
> there's too much for us to lose at this point Feeling entitled?
907.
by jampekka 2y ago
That typically means two totally different implementations, and pure Python versions are often unusably slow, so it doesn't help much to hack that.
908.
by jampekka 2y ago
JS array access is a lot faster than Python array access. JS is easily within magnitude of C and can be even about as fast with typed arrays or well JITable code.
909.
by jampekka 2y ago
Python is really really slow compared to JS though.
910.
by jampekka 2y ago
Hopefully the lot includes that writing stuff in low level languages isn't worth the pain most of the time.
911.
by jampekka 2y ago
You can experience this easily yourself: https://en.m.wikipedia.org/wiki/Motion_aftereffect Motion is separated from shapes and structure and objects changing place very early in visual processing, and it's kind o
912.
by jampekka 2y ago
I'd guess few people have much image-like things in their head for data structures or control flows or such. Such can be conjured if one tries to explain something visually by e.g. drawing. But 2D/3D structure is very limited for
913.
by jampekka 2y ago
So the plan is to make AI not-evil by doing it with Microsoft and Oracle?
914.
by jampekka 2y ago
If anything is dying it's proprietary software. Which is great for all of us because open source is vastly more efficient system.
915.
by jampekka 2y ago
Not bad, but what's up with the audience? Is there an Adobe cult or something?
916.
by jampekka 2y ago
No documents needed for a prepay number. And no smartphone required for banks, at least all of them. Authentication can be done with a password and a list of passcodes from the bank.
917.
by jampekka 2y ago
At least in Finland not having an address is not a huge hassle. You can always put poste restante (a post office pickup) address if needed. Physical mail is quite rare for official business, and almost everything can be done online. But not
918.
by jampekka 2y ago
Not exactly 3D assets, but diffusion modems are used to generate e.g. traffic (vehicle trajectories) for evaluating autonomous vehicle algorithms. These vehicles tend to crash quite a lot. For example https://github.com/NVla
919.
by jampekka 2y ago
It is used as a predictive model of the environment for model-based RL. I.e. agents can predict consequences of their actions.
920.
by jampekka 2y ago
It took EU enforcement eight years to enforce obvious violations of the law for a few companies. The illegal nags are still rampant. And e.g. Meta (and many newspapers) already has a new obviously illegal tracking scheme with the "pay
921.
by jampekka 2y ago
It's not a stupid opinion to have different productivity/risk tradeoffs. A lot of "all admins" don't seem to care much about productivity, yank and unintended consequences. "All admins" pushed the ridiculo
922.
by jampekka 2y ago
Surely they can't be thinking much about operating systems?
923.
by jampekka 2y ago
The law is not enforced. The non-enforcement is largely by design/lobby though.
924.
Archive.org Hacked?
6 points by jampekka 2y ago|2 comments
925.
by jampekka 2y ago
Tangential: Archive.org is giving alert popup "Have you ever felt like the Internet Archive runs on sticks and is constantly on the verge of suffering a catastrophic security breach? It just happened. See 31 million of you on HIBP!&quo
926.
by jampekka 2y ago
And medicine and literature and chemistry and war. Maybe they'll get all the prizes?
927.
by jampekka 2y ago
TBF backpropagation was introduced only in the 1970's, although in hindsight it's a quite trivial application of the chain rule. There were also plenty of "hacks" involved to make the networks scale such as dropout regul
928.
by jampekka 2y ago
There were some predictions that Peter Shor could win this year for quantum computation. I'd say his work is a lot closer to physics than Hinton's or Hopfield's.
929.
by jampekka 2y ago
Neither Hopfield networks nor Restricted Boltzmann Machines/Deep Belief Networks really panned out for any purposes outside some theoretical niches. The prize was awarded for "AI" and the tenous links to physics of some irrel
930.
by jampekka 2y ago
Neither of these models were never really influential though beyond some theoretical niches.

More ›