0xbear

45 ms·

421.
by 0xbear 9y ago
True story: in Russia (and I can only assume other Eastern European countries) peanut allergy is so rare that I've never even heard of it until I emigrated. Pollen allergy is about the same, ragweed pollen allergy can be really bad too
422.
by 0xbear 9y ago
Well, the manufacturing order is probably for a few tens of thousands, and if sales don't go as they hoped, they could cut the price. As far as Android phones go it's not a bad phone. But pricing it at the iPhone levels was a mist
423.
by 0xbear 9y ago
Now we'll know exactly how many Andy Rubin fanboys there are in the world. Why would anyone else buy a $700+ phone from a startup which might not even be there next year? My prediction: a few tens of thousand units will be sold mostly
424.
by 0xbear 9y ago
If you don't round up at least 2 offers at the same time, you're 100% getting ripped off in comp negotiation. I _hate_ interviews, but I go the extra mile to get that second offer, because that's how this game is set up.
425.
by 0xbear 9y ago
Large datasets will still be important for all but strictly perceptual tasks.
426.
by 0xbear 9y ago
The largest ad system ever, Google AdSense, used MySQL until circa 2014 when it moved to a completely custom DB backend, F1. F1 is also a SQL database, however. Google does use bigtable and such where appropriate, but for anything more comp
427.
by 0xbear 9y ago
>> For example, students and professors I met in college >> that grew up in the USSR thought engineering was >> stereotypically women’s work Um, no. I grew up in the USSR, too, and engineering never was stereotypically w
428.
by 0xbear 9y ago
They do have "stakeholders" which, naturally, can't profit from either operation or sale of the nonprofit, but, crucially, _can_ control creation of commercial spin-offs and spending. I.e. While de jure they can't be cal
429.
by 0xbear 9y ago
That's a common misconception. Nonprofits _do_ have owners. And also, while they can't make money directly, they can spin off for-profit companies if e.g. they invent a cure for cancer or something.
430.
by 0xbear 9y ago
About 15-20% of full strength. Hardly negligible.
431.
by 0xbear 9y ago
Decaf does contain caffeine though.
432.
by 0xbear 9y ago
I wouldn't be quite so quick to jump to the "willful" label. I'm not a native, so I know very little about US Civil War. I also know that much of what you think you know about the war is actually bare propaganda written
433.
by 0xbear 9y ago
Why stop there, though? There were _many_ US historical figures who owned slaves, and more, were in favor of slavery. If we're going to be principled about it, let's start by renaming Washington DC and go on from there all the way
434.
by 0xbear 9y ago
Dude himself has been doxxed by 4chan already, and confirmed his identity on TV. I figure that twitter account has at most couple of weeks of life left in it, until the wrongfully accused serve the guy with legal paperwork. Just to save you
435.
by 0xbear 9y ago
That's still almost $5.5 million. Less than the cost of development, but probably not that far off.
436.
by 0xbear 9y ago
Is Mono ready for the prime time? Because it's most certainly already in Debian. It's just that MS doesn't give a shit about adoption on the "cancer" platform. They'd much rather we all wore the Windows ball an
437.
by 0xbear 9y ago
PPA is specifically designed for packages that are potentially not stable. And you can find dozens of garbage packages in Debian repo without even trying all that hard.
438.
by 0xbear 9y ago
Still no Ubuntu PPA or Debian upstream packages. Apparently they don't want people to actually use it.
439.
by 0xbear 9y ago
Are sure it's me and not you? I have a 13 year old kid, and he still can't figure out that if he doesn't turn in his homework on time his grades will suck. :-)
440.
by 0xbear 9y ago
Why do children under 8 need to know how to code? They don't even have simple things like cause and effect established in their brain until they're about 11. I.e. "if it's raining outside I should wear a rain coat or I&#
441.
by 0xbear 9y ago
We just plain can't do data augmentation quickly enough with TF. Queues-schmeyes, doesn't matter. Still tops out at about 35MB/sec on MS COCO and starves even a single Titan Xp. On the same hardware, with the same data augmen
442.
by 0xbear 9y ago
So was the reply. I didn't know about that other study. I also have no reason to believe it is more (or less) authoritative than the NASA study. Most people don't actually know that the Antarctic (unlike the Arctic) ice sheet is n
443.
by 0xbear 9y ago
There is always "doubt" in science. That's why it's science and not religious dogma. Who was right, however, still remains to be determined.
444.
by 0xbear 9y ago
Yeah, try to open those WYSIWYG editor docs, say, 20 years from now, and let us know how well this worked out for you.
445.
by 0xbear 9y ago
Except, of course, _Antarctic_ ice sheet just keeps growing. https://www.nasa.gov/feature/goddard/nasa-study-mass-gains-o...
446.
by 0xbear 9y ago
Why the downvotes? V100 does have a TPU-like unit in it.
447.
by 0xbear 9y ago
I worked with it _before_ it was released, and with DistBelief before that. For a long time I considered TF to be the best option for the class of problems I work on. I was also extremely skeptical of PyTorch, but it won me over. Now that I
448.
by 0xbear 9y ago
TPU will be everywhere next year in the form of NVIDIA Tesla V100. And FB is clearly getting much better ROI. It's not even close.
449.
by 0xbear 9y ago
Take a look at PyTorch. Much easier to understand, fewer bugs, much less obscure all around. I've spent years working with TF and have come to dislike its inflexible, verbose, and convoluted way of doing things. I'm 80% on PyTorch
450.
by 0xbear 9y ago
I've tried XLA and observed no measurable benefit for most models, small benefit for some, and worse performance for others. That was a few months ago, so maybe things have changed since then. It won't be a huge difference no matt

More ›