Veedrac

42 ms·

511.
by Veedrac 6y ago
Progress on online learning builds on progress on offline learning, because it's the same optimization procedure. I don't really agree with your stance on it, but I was not objecting to the claim that people should focus more on
512.
by Veedrac 6y ago
There is plenty of research on exactly this sort of learning-while-doing, termed ‘online learning’. It's easiest to train on IID data, and also common to train in specific RL environments, because those paper over some issues with cata
513.
by Veedrac 6y ago
I disagree, but that wasn't the argument I was objecting to.
514.
by Veedrac 6y ago
> The fact that you don't consider ants as general intelligences means you agree that we aren't making progress towards general intelligences, since ants intelligence is way closer to general intelligence than any computer prog
515.
by Veedrac 6y ago
Our computers also don't know how to do all sorts of complex biological manufacturing that small groups of proteins can do inside cells. Thus your measure of intelligence also says that protein complexes are intelligent, but they are
516.
by Veedrac 6y ago
So therefore protein complexes are intelligent? This is literally the watchmaker argument with the nouns swapped out.
517.
by Veedrac 6y ago
> Our super computers can't even compete with the intelligence of an ant I don't see what might spur anyone to claim this. It's obviously ridiculous.
518.
by Veedrac 6y ago
This is what I've started thinking of as the ‘pretty dumb for a human’ take on AI. If your requirement for believing something is on the path to AGI is for it to practically already be AGI, it's no longer predictively meaningful.
519.
by Veedrac 6y ago
This is a really dangerous line of argument for an example where one platform has a de-facto monopoly on views, whose censorship is fine-grained enough that people open to these other ideas simply won't be able to find them. It makes a
520.
by Veedrac 6y ago
Censorship does work, that's why countries do it.
521.
by Veedrac 6y ago
Yes, since 2007.
522.
by Veedrac 6y ago
The first page was leaked. The environmental angle was a significant part of it, particularly the claim that environmental and financial costs ‘doubly punishes marginalized communities’.
523.
by Veedrac 6y ago
> We end with recommendations including weighing the environmental and financial costs first ‘Train for less time,’ > investing resources into curating and carefully documenting datasets rather than ingesting everything on the web ‘on
524.
by Veedrac 6y ago
> it is incredibly anodyne I don't get why anyone is surprised that Google doesn't want a paper saying ‘we should spend less money on ML’ published under their name, when the whole company is ML driven and pushing for growth th
525.
by Veedrac 6y ago
> e: although AlphaFold appears to be convolutionally based! I suspect that'll change soon. “For the latest version of AlphaFold, used at CASP14, we created an attention-based neural network system” ?
526.
by Veedrac 6y ago
If anything Apple has proven that the OoO power tax is mostly imaginary, so we'll probably just get more, bigger OoO for a while.
527.
by Veedrac 6y ago
The true answer is we don't know what Apple's scaled-up GPUs will look like, so it's hard to tell precisely how performant they will be. However, everything so far indicates these will be pretty powerful, as even the M1 is pr
528.
by Veedrac 6y ago
No, they're aiming at training too.
529.
by Veedrac 6y ago
> It just says "oh, it's a hardware toggle"; toggle of what exactly? A toggle that makes the chip treat all loads and stores from that thread as TSO.
530.
by Veedrac 6y ago
Give it a few months, since the Cortex X1 should be competitive in laptops.
531.
by Veedrac 6y ago
Apple is about platform control. Apple Silicon is about performance and efficiency. A friendly, open-hardware Apple would have moved to their own chips too, though they'd have documented their GPU, and a walled-garden Apple without
532.
by Veedrac 6y ago
> "Calm" me? A poor choice of words, sorry. I meant, I understood you to be saying you found the criticism of Louise unpleasant, and I thought it would lessen that to know that I didn't and don't think Louise was bad.
533.
by Veedrac 6y ago
> I believe I have remained polite and respectful towards you throughout I disagree. I do not mind in the slightest being told I am wrong, or having my ideas criticized. But calling me too stupid to understand my own point, or too intell
534.
by Veedrac 6y ago
As I said, rather don't reply than reply unpleasantly. I'm cutting this here.
535.
by Veedrac 6y ago
> Louise's biases are not problem-specific in the short example I showed you. This is clearly untrue. You were customizing the BK to each specific task. You were also customizing the stepping stones for each specific task. Justifica
536.
by Veedrac 6y ago
My initial response was a fairly kneejerk reaction to the snark. The following is a rewrite. Please don't; if you really think so little of me, rather don't reply than reply unpleasantly. > Deep neural nets do not learn only fr
537.
by Veedrac 6y ago
Nobody wanted Arm-based Windows devices. The early RT devices had terrible performance and failed to compete in any meaningful way, had terrible compatibility and no (?) emulation, and since the products were mostly terrible, Microsoft neve
538.
by Veedrac 6y ago
> I mean, try to define some BK and generate combinations of it until you solve the problem. You say that I made the problem easy because I defined some BK by hand. I suggested you try doing that to see whether the problem is as easy as
539.
by Veedrac 6y ago
> in the 16-inch MacBook And the 13". We'll probably have something even better for the 16" too.
540.
by Veedrac 6y ago
HN seems to hide replies once they get deep enough. I honestly think we've mostly exhausted that conversation, though.

More ›