Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
phoe-krk
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
28 ms
·
361.
▲
by
phoe-krk
4y ago
Yes, agreed, I don't think the problem is with networks that mix tons of input data in a way that doesn't heavily derive from one or a couple of sources. The currently available models do not have overfitting solved, though, and t
362.
▲
by
phoe-krk
4y ago
This might be technically correct but doesn't seem to matter much in practice because of how many practical cases of NN outputting copyright-infriging content there are. These include examples in the recent lawsuits that made the round
363.
▲
by
phoe-krk
4y ago
If you have an image, then train a neural network on that image, then use the neural network to reconstruct that image in detail, then the NN by definition contains enough information used to reconstruct that image - hence, a copy. With NNs
364.
▲
by
phoe-krk
4y ago
> But the latest and greatest software trend–generative AI–is in danger of being swallowed up by copyright law. About time. > If the AI industry is to survive, we need a clear legal rule that neural networks, and the outputs they p
365.
▲
by
phoe-krk
4y ago
I assume that's one of the parts of the work done at https://github.com/WebAssembly/gc - not happening any soon yet, but it'll eventually be done.
366.
▲
by
phoe-krk
4y ago
I meant that automatic garbage collection is not always the way to go for some people, especially in the days when JVM GCs were not as advanced as they are now.
367.
▲
by
phoe-krk
4y ago
Also the Common Lisp OS Mezzano[0] runs like that. It runs Doom and Quake by means of lowering C to LLVM-IR which is then lowered to a subset of CL which is then compiled by the trusted native Mezzano compiler. [0] https://github
368.
▲
by
phoe-krk
4y ago
Not until hardware became fast enough and compilers performant enough to support this, and not until everyone had to learn their lessons with approaches like the JVM.
369.
▲
by
phoe-krk
4y ago
Yes. Probably the most scary thing is that this part is becoming more and more true as well.
370.
▲
by
phoe-krk
4y ago
Yes, the video is pretty amazing - Gary in 2014 managed to predict pretty much everything that's happening with WASM nowadays: the push for safe execution of software compiled in unsafe languages, a common VMesque instruction set, web
371.
▲
by
phoe-krk
4y ago
Your response is technically correct but misses the entire point of the video. To make a tl;dr, the video talks about a technology called asm.js which other applications end up compiled into so they can run with near-native speeds on any sy
372.
▲
by
phoe-krk
4y ago
It's amazing to see https://www.destroyallsoftware.com/talks/the-birth-and-death... becoming more and more true with every year.
373.
▲
by
phoe-krk
4y ago
> I wonder how many people's RSS/JSONFeed readers we broke with the title... Well played.
374.
▲
by
phoe-krk
4y ago
I think the mistake is to consider "human" something separate from the rest of the world as opposed to a more holistic point of view. The fact that we have a somewhat complicated mind capable of abstract thinking doesn't mean
375.
▲
by
phoe-krk
4y ago
Sure, the universe seems infinite enough both in time and space. Assuming that laws of statistics hold, there should be other areas where there's optimum for life as we know it to develop - even if we won't get to meet it because
376.
▲
by
phoe-krk
4y ago
> One obvious objection to neutrality is the threat of extinction. If one couple refuses to have a child, it is neither good nor bad. But if every couple refuses, it is a catastrophe. If the last human dies and there is no one there to
377.
▲
by
phoe-krk
4y ago
These badges are emoji - currently e.g. https://infosec.exchange/@lcamtuf has three of them. That's 200% more than you can get on Twitter, and $8 cheaper, too. Actual verification happens via proving that you control a
378.
▲
by
phoe-krk
4y ago
Email mailing groups figured out a way of scaling despite the same sort of exponential congestion. If anything, Fediverse traffic between larger clusters may eventually ossify around some larger pipes, not unlike the Internet traffic being
379.
▲
by
phoe-krk
4y ago
Donating to server admins seems to work for the instances whose growth I'm in/directly observing. And then, running an instance with a few hundred users is not a resource hog.
380.
▲
by
phoe-krk
4y ago
Yes, and we're already seeing it e.g. with https://mas.to getting hugged to death because of pg joining it. If anything, it can be treated as a technological problem, solvable e.g. by (theorizing now) splitting a single ins
381.
▲
by
phoe-krk
4y ago
> Mastodon instances do not have "terms of service" they have "rules". The "rules" are unilateral in that they bind the user and not the service provider. Of course they do. If the instance administrators
382.
▲
by
phoe-krk
4y ago
> And, on the non-technical side, it has a disorganized anonymous set of moderators you've never met or ever heard of, with no particular guiding philosophy or rules, and these guys have total access to your profile and they can do
383.
▲
by
phoe-krk
4y ago
It's an error to treat the Fediverse as you would treat Twitter, where there is a singular platform in some way external to its users. Fediverse is a mesh network with single-user instances being feasible. If a single node has problems
384.
▲
by
phoe-krk
4y ago
Direct Fediverse link: https://mas.to/@paulg
385.
▲
by
phoe-krk
4y ago
His hyping/scamming: is it good for Tesla, or is it bad for everyone else though?
386.
▲
by
phoe-krk
4y ago
> I'm sure those people would rather see Tesla go down due to its failures, not because of Twitter. Tesla's most important failure is Musk. That's why TSLA shareholders are trying hard to oust him.
387.
▲
by
phoe-krk
4y ago
Possible dupe: https://news.ycombinator.com/item?id=34010112
388.
▲
by
phoe-krk
4y ago
I don't think you need to reach infinity to realize it's OTCA metapixels all the way down and OTCA metapixels all the way up. In a way, it just loops forever.
389.
▲
by
phoe-krk
4y ago
> What does it look like if you zoom out infinitely often? Make a sharp backwards roll of your mouse wheel and see for yourself - it'll turn on the autopilot mode.
390.
▲
by
phoe-krk
4y ago
> That would make me lower my trust of that author. Would it? How would you tell? How can you prove that a piece of text was not written by an artificial intelligence? > (likely because the author is human) I think that's t
More ›