blt

45 ms·

511.
by blt 6y ago
I agree with the main idea of your argument, but how can it be set in the "space of all differentiable linear functions" when trigonometric functions aren't linear? I think the argument would hold on the space of real analyti
512.
by blt 6y ago
Re: the "what is ROS?" question asked by many on this thread: Like others have said, ROS includes a serialization protocol with codegen, a pub/sub and RPC messaging system with discovery, a build system, a configuration file
513.
by blt 6y ago
They still make exciting technology. Google's search is still far better than the competition . The Pixel computational photography app is cutting edge. Their translation and speech synthesis are great. They are still good at "or
514.
by blt 6y ago
Classically, the equivalent of complexity theory in machine learning is statistical learning theory, where the main question is: if I have a magical algorithm that always finds the function in my class that fits the data best, how big does
515.
by blt 6y ago
This is a great resource. I will return to read it closely next time I am designing a CLI. One thing that puzzled me: `git push` is given as an example of the principle "If you change state, tell the user". Its output is: $
516.
by blt 6y ago
Laypersons often use the word "solution" in situations where an academic would say "method" or "approach": we did something useful, but it may not be the best possible way. In pure math, "solution" me
517.
by blt 6y ago
The bit manipulation already generates an approximation of 1/sqrt(x), not just 1/x.
518.
by blt 6y ago
Scraping images is not enough. There's still pagination to deal with. It's also important to scrape the post text and comments.
519.
by blt 6y ago
Facebook actively wants to make this hard. I hope you figure out a way. If so, please post to HN. I think many Facebook group users and admins would be interested.
520.
by blt 6y ago
I am glad non-destructive editing is on the roadmap. Hopefully we'll have it within the next 25 years. Photoshop had adjustment layers in 1996.
521.
by blt 6y ago
I did draw solid regions.
522.
by blt 6y ago
It is very much not working for me. I get a mostly-transparent creature. The opaque parts seem to follow the brushstrokes of my drawing. Maybe it is because it was trained on finely detailed material, whereas my drawing was a coarse doodle.
523.
by blt 6y ago
I agree that vim makes you much more line-oriented, but I'm not sure it's always a good thing. My non-vim collaborators find it strange that my LaTeX files have each clause of a sentence in a separate line.
524.
by blt 6y ago
Your first statement assumes that linear programming solvers can be massively parallelized. Can they?
525.
by blt 6y ago
What are those Rolls Royce features needed by chauffeurs?
526.
by blt 6y ago
We should not be surprised to see new language models achieving GPT-3 performance with fewer parameters. The purpose (I assume, as an outsider) of the GPT-* project is to try to find the upper bound of how good a language model can be, with
527.
by blt 6y ago
OOP is a nice feature for a language to have. The backlash was against the religious fervor around OOP. People were claiming that OOP is the one true way to organize software. Java's designers went so far as to eliminate the concept
528.
by blt 6y ago
X-plane is also pretty bad for this. It has VOR and ILS tutorials, but nothing more advanced navigationally (like navigating to the intersection of two VOR radials), nothing about crosswind landing, nothing for the more complex aircraft, et
529.
by blt 6y ago
SWIG can sometimes help with this.
530.
by blt 6y ago
So they don't make their own power supplies? Who is making them? Is there a better power supply brand that manufactures their own?
531.
by blt 6y ago
yes, I get this too.
532.
by blt 6y ago
I participated in MetaFilter for years before I went to any meetup. At my first meetup, I realized I had never thought about how to pronounce the nickname "MeFi" despite reading and typing it countless times.
533.
by blt 6y ago
Chinese people don't use US social media networks.
534.
by blt 6y ago
yes, they are conflating "eigenvector of the covariance matrix of a probability distribution" with the general concept of an eigenvector.
535.
by blt 6y ago
This insight is the essence of the AlphaZero architecture. Whereas a pure Monte Carlo Tree Search (MCTS) starts each node in the search tree with a uniform distribution over actions, AlphaZero trains a neural network to observe the game sta
536.
by blt 6y ago
I agree. Just like other old things that are still around, old art seems good due to survivorship bias. There has been lots of terrible art made in all media since the beginning of civilization.
537.
by blt 6y ago
Usually people who say things like this have made zero effort to discover contemporary art that they like. You list examples of old European art that can be found in a 7th grade textbook.
538.
by blt 6y ago
Those samples were probably written by someone who had learned how to code in assembly language. In assembly, this kind of comment reminds the reader of the semantic meaning of what is in each CPU register. It could be very useful. Then the
539.
by blt 6y ago
I think the limiting factor of the color intensity is the dye/ink used to make the mask. Probably each pixel is only blocking like 50% of the light in its stop band.
540.
by blt 6y ago
Yeah, irregularly spaced time series data seemed to be the killer app of Neural ODEs according to my understanding.

More ›