Eridrus

35 ms·

541.
by Eridrus 6y ago
It's hard to find out how much has been invested in self driving cars, but I think you're massively underestimating the cost of building rail in the US. A billion dollars a mile is a realistic figure for rail in cities. NYC spent
542.
by Eridrus 6y ago
I would try the account recovery process: https://support.google.com/accounts/answer/7682439?hl=en
543.
by Eridrus 6y ago
Is it informative? Where did these numbers come from? Why are they good thresholds? As someone living in NYC, getting to some threshold Cuomo pulled out of his ass doesn't mean anything to me.
544.
by Eridrus 6y ago
I feel.lioe so much of this discussion misses that all of these things about people are both situational and mutable. I've done some WFH on every job I've had, with wildly different results, depending on how motivated I was to do
545.
by Eridrus 6y ago
I think these pruning methods do work if you put some effort into engineering them properly. But I came here to agree about most researchers having no interest in actual inference time performance. I just tried a library that was meant to b
546.
by Eridrus 6y ago
Your models might be ridiculously tiny, but a lot of people's models are not. Take a loot at any research paper in vision, speech or language. The models are gigantic.
547.
by Eridrus 6y ago
Having seen a bunch of people fail to make distillation work in the sense of training a better model than by training the smaller model from scratch, I don't think it's really fool proof...
548.
by Eridrus 6y ago
Quantization is orthogonal and complementary.
549.
by Eridrus 6y ago
It should be pretty obvious why, they don't work as well as what is standard. This is about ways to explain the models we get good performance with.
550.
by Eridrus 6y ago
Maybe Schmidt really expects a war with China, but I find it hard to be impressed by an XX% efficiency gain in finding potential terrorists. Obviously these problems are difficult in some way, otherwise they would have been solved, but at l
551.
by Eridrus 6y ago
I'm not an idealist over here, but It's really weird to me that when he looks at all of the US government's tech problems, he singles out the military as the part that it is really important to improve. But I guess it doesn&#
552.
by Eridrus 7y ago
This doesn't address your comment about device users per se, but you can have multiple accounts on your device and meet will let you choose which one to use when joining calls.
553.
by Eridrus 7y ago
For a hobbyist, you definitely don't want to use the cloud, buy yourself a good GPU (I got a great deal 2nd hand in eBay) and you'll save a lot of cash. What you need the cloud for is parallelism. If you want to do any sort of hyp
554.
by Eridrus 7y ago
Yeah, the gif is terrible... the AV1 side is on the left.... https://www.blog.google/products/duo/4-new-google-duo-featur...
555.
by Eridrus 7y ago
> They couldn't explain it if they tried - nobody could - it's all generated by stochastic ML methods. This isn't true. These systems are not unknowable. It's just generally not worthwhile to figure out what these enc
556.
by Eridrus 7y ago
The reason people say not to roll your own crypto is that there is no secret answer to making things secure, we just have smart and creative people bash their heads against a crypto protocol/implement for a long time and hope we found
557.
by Eridrus 7y ago
Any bit of reduced transmission helps. All we need is for R0<<1 and then this pandemic will end. Any contacts this finds that human contact tracers do not is a win.
558.
by Eridrus 7y ago
Honestly, this was foreseen by everybody writing commentary. All the coverage of the paycheck protection program mentioned that the loan terms were extremely generous, and the money was not enough to cover everyone who was eligible, so it
559.
by Eridrus 7y ago
No, because not all jurors are equally easy to intimidate, so you should expect effort to be superlinear depending on the distribution, and intimidation itself is not a risk free crime.
560.
by Eridrus 7y ago
I was not arguing that this is the "best it can get", just that there is no structural answer to these challenges. There's no way to setup our institutions such that they cannot be influenced by interest groups, so the only t
561.
by Eridrus 7y ago
One interesting funding mechanisms in this space is Emergent Ventures / Fast Grants. It functions more as a non-profit than a VC entity, including recently organizing $10m in funding for Covid-19 research, including from a bunch of nam
562.
by Eridrus 7y ago
This isn't just a big business problem, there is plenty of regulatory capture on behalf of other interest groups (eg doctors keep the supply of new doctors low). There is no enduring structural answer to this problem, it is fundamental
563.
by Eridrus 7y ago
I have a Linux laptop from work with a small amount of supported hardware configurations. I have had issues with: UI lockups, audio failures, resume from suspend issues. Windows 10 isn't perfect, but it's a lot better at not getti
564.
by Eridrus 7y ago
I appreciate you trying to make more content beyond the usual intro content more accessible, but if you really are trying to share things in an accessible manner, you probably need to lose the "Get used to it" attitude that you wr
565.
by Eridrus 7y ago
Yes, but a forum where you would expect the quality of discussion to be high. I feel some sympathy for grad students, but am personally quite excited to see how this plays out since I've wanted virtual conferences to be more of a thing
566.
by Eridrus 7y ago
Maybe you can file this under hating journalists, but it seems like a culturally and politically influential industry trying to extract money from everyone else.
567.
by Eridrus 7y ago
a) It's about letting people resume their activities and keep a low R0. b) It's not 0, so there's still room to reduce the number, even now.
568.
by Eridrus 7y ago
The Apple/Google design drops this DP-3T requirement: 2) Enable epidemiologists to analyse the spread of SARS-CoV-2 So anything in that table with epidemiologists is gone. The remaining caveats are pretty boring: To do so, the attacker
569.
by Eridrus 7y ago
I don't know how effective this was, but Israel did exactly this: https://techcrunch.com/2020/03/18/israel-passes-emergency-la... US public institutions seem frankly sclerotic. The fact that the governme
570.
by Eridrus 7y ago
If you assume the goal of contact tracing is to literally find every infection, yes, this isn't going to work. If you assume the point of this is to reduce R0, then this will work just fine at any stage of the pandemic. There's ob

More ›