ladberg

51 ms·

541.
by ladberg 6y ago
The memory model of the ISA != the memory model of your code. C has a memory model and any correct C code is compiled to the correct ARM assembly. It's only really an issue if you are translating over x86 assembly, and then you probabl
542.
by ladberg 6y ago
"Defund the police" doesn't mean taking away all police. It just means reducing funding and reducing responsibilities of the police to avoid them becoming a catch-all for any civil issue.
543.
by ladberg 6y ago
I feel like there's a ton of unexplored area in the hardware space for extremely low power circuits used for ML. As we know, you can do ML with very low precision, which is why chips that can do fast INT8 and F16 are getting more popul
544.
by ladberg 6y ago
I think it's just a sieve that removes multiples of 2 and 3, leaving false positives that are multiples of 5, 7, 11, etc.
545.
by ladberg 6y ago
I feel like more and more transistors will be dedicated to machine learning tasks. You'll eventually have phones and computers constantly running neural networks to accomplish advanced but relatively mundane tasks (e.g. text-to-speech,
546.
by ladberg 6y ago
If you turn Javascript off, the only information the website can get is user agent and IP, which would narrow it down much less than using Javascript even just among the pool of non-javascript users. Keep in mind that there are a lot of ser
547.
by ladberg 6y ago
Is there any application that will benefit from FHE? I can't think of anything where taking a hit of at least 5 orders of magnitude is worth it over doing the calculation locally on a trusted CPU.
548.
by ladberg 6y ago
If you know the exact position and direction of a photon coming out of the accretion disk, you could theoretically reverse out a trajectory and calculate what direction it approached the black hole from at what time. Of course, that relies
549.
by ladberg 6y ago
I think it's all point-to-point directional antennas, so it would be linear.
550.
by ladberg 6y ago
The author mentions that the shape of the curve from 117M parameters to 175B is interesting, but doesn't show it. Does anyone have the graph?
551.
by ladberg 6y ago
Does anyone know of something similar for non-ML papers?
552.
by ladberg 6y ago
Pretty cool, but it also looks like it relies on having an incredibly accurate map of the city. Does anyone know if there are autonomous driving systems that don't need that detailed of a map (e.g. just using Google Maps)? There's
553.
by ladberg 6y ago
I loved Nitrome when I was younger and I am super excited for this! I haven't visited in years mainly because I don't have Flash.
554.
by ladberg 6y ago
Oh yep I was doing light sampling and I skimmed the article so I didn't realize they weren't. Thanks!
555.
by ladberg 6y ago
I don't mind, I never used them and I feel like most people I know didn't either. That said, they're sold separately for $29 which seems a bit steep, so hopefully they'll get a price drop for the people who do want them.
556.
by ladberg 6y ago
Very cool! I'm a bit curious about where the performance losses are. I just made a CPU path tracer for a graphics class and was able to render similar scenes in much less time. This is running on a GPU so it should be much faster, but
557.
by ladberg 6y ago
Yep, that's the point of FPGAs. The easiest way to see is to synthesize it in an FPGA tool (e.g. Vivado) which you can probably do for free. Different FPGAs have different hardware resources, but for any small designs you can probably
558.
by ladberg 6y ago
I could be wrong, but I feel like deep learning denoising is mostly just blurring in the right places. Deep learning is good at segmenting each part of the image that's supposed to be a different color, and then denoising can be done b
559.
Show HN: Turn an image into Unknown Pleasures style art(github.com)
6 points by ladberg 6y ago|0 comments
560.
by ladberg 6y ago
Very cool looking, but the marketing is a bit misleading because I feel like you'd have to go back 10 years to find a 15" laptop that big.
561.
by ladberg 6y ago
Pretty much everyone has already seen the examples of interpolating between two faces, so I really like that they show what happens when you interpolate PAST a face. For example, they make an image of person C that is derived from person A
562.
by ladberg 6y ago
Yep, but not as well as tech internships. Still, the company was throwing money down the drain because they didn't know that even a semi-competent programmer could automate those jobs.
563.
by ladberg 6y ago
I get that Roger Penrose is an incredibly smart guy, but I really don't see how he comes to the conclusion that there are any quantum effects in the brain. Even if there are, I don't see how they are necessary in any way for consc
564.
by ladberg 6y ago
Not OP but I had a friend who has interned at a few record labels. He's not really a programmer, but knows enough about programming to know what kind of task should be automated. He described a portion of his job to me that he and a bu
565.
by ladberg 6y ago
It only destroys the goods if you let them soak in the solution for hours. How does that make it dangerous? If someone wanted to just destroy goods, fire is way more effective.
566.
by ladberg 6y ago
I would also add IMO the main reason that Java matters: it's easily moddable without the creator's permission. For most of Minecraft's existence, there was zero official mod support and zero in-game scripting abilities. Howev
567.
by ladberg 6y ago
FYI, it seems more complicated than just precedence: >>> True == False is False False >>> True == (False is False) True >>> (True == False) is False True
568.
by ladberg 7y ago
Most "satellite view" imagery of cities is actually done with planes and drones. I just looked up a 3D model of my house in Apple Maps and can see stuff with a few centimeter resolution, which is much higher than any satellite can
569.
by ladberg 7y ago
Kind of an interesting tangent, but the founder of Lytro (Ren Ng) is the advisor on a really cool paper that came out recently and reminds me a lot of this: http://www.matthewtancik.com/nerf Basically, it's similar to
570.
by ladberg 7y ago
If you like this then you'll love neural radiance fields: http://www.matthewtancik.com/nerf

More ›