programjames

139 ms·

211.
by programjames 1y ago
There is, and people have trained purely optical neural networks: https://arxiv.org/abs/2208.01623 The real issue is trying to backpropagate those nonlinear optics. You need a second nonlinear optical component that
212.
by programjames 1y ago
Why not just subsume it into the other income taxes?
213.
by programjames 1y ago
Tuition money doesn't fund research, so how is the GI Bill relevant?
214.
by programjames 1y ago
I feel like the math term closest to that would be "consistent" not "efficient".
215.
by programjames 1y ago
Why do finance bros like to call allocating all the resources to the maximal-dollar extracting process "efficient". It's not. All they're doing is turning the economy's entropy into dollar bills now, which stunts fu
216.
by programjames 1y ago
Carcinisation in action: free (foss) -> non-profit -> capped-profit -> public benefits corporation -> (you guessed it)
217.
by programjames 1y ago
You can train an LLM to maximize the information content bitrate. I just think most companies want to maximize "customer satisfaction" or w/e, which is why we get the verbose, bold, bullet points.
218.
by programjames 1y ago
I can only find papers from the late 1980s on this, but a professor around during the 60s told me hackers were working on robotic ping-pongers way back then.
219.
by programjames 1y ago
People did the conveyor belt thing in the 60s.
220.
by programjames 1y ago
Feels like clickbait. People were doing similar things in the 60s at MIT. I thought by "human level robot" they meant actually human level , with a humanoid robot swinging the paddle, not a conveyor belt.
221.
by programjames 1y ago
As someone born in 2003, I did this just last week when filing my tax returns.
222.
by programjames 1y ago
Sure. The original normalizing flows used a fixed number of layers. Someone at UToronto recognized that, as the number of layers gets very deep, this is essentially an ordinary differential equation (ODE). Why? Suppose you have n residual l
223.
by programjames 1y ago
This is why I believe a numerical methods course should be a requirement for any AI majors.
224.
by programjames 1y ago
Problem 7 here: https://mathcontest.unm.edu/PastContests/2016-2017/2016-2017... Or a Veritasium video here: https://www.youtube.com/watch?v=iSNsgj1OCLA
225.
by programjames 1y ago
69% looks surprising like the answer to this puzzle: > The numbers 1–n are randomly placed into n boxes in a line. There are n people who are each able to look into half the boxes. While they are allowed to coordinate who looks into whic
226.
by programjames 1y ago
It seems absolutely bonkers to me that someone would write a smart contract that lets them bleed $50m without automatically stopping after they lose the first $1/10/100k.
227.
by programjames 1y ago
Next we're going to learn that winning Poker Bots with an "all in" strategy is defrauding the competition.
228.
by programjames 1y ago
I thought there was a little more structure than just that (since that's been around since the 1940s).
229.
by programjames 1y ago
Here's a link to the paper: https://www.nature.com/articles/s41586-024-08262-7 I really wish they tied this into representation theory, because it looks like the R-tensors are similar to representations of the p
230.
by programjames 1y ago
> Something that surprisingly few people have mentioned recently is that having the biggest trade deficit is the same thing as being the global reserve currency. Can you explain this a little more? Why does having a trade deficit make yo
231.
by programjames 1y ago
I'm a constructivist, but I still believe in proof by contradiction. In fact, I don't think you can believe in proof by contradiction without constructivism. How do you know you can take an arbitrary proof P, and show it leads t
232.
by programjames 1y ago
I consider it false advertising to not tell me the drawbacks of your product compared to your competitors. I shouldn't have to look up every competitor to find your car is the only one without a steering wheel, even if it has heated
233.
by programjames 1y ago
> No money is exchanged when I promote my own product through my own channels, nor when I promote my friends products, whether I disclose it as promotion, or disguise it as my genuine unaffiliated opinion. But this is not the vast majori
234.
by programjames 1y ago
Well, I think the point is your friend can still tell you he likes his new pair of pants, he just can't be a paid actor.
235.
by programjames 1y ago
I think the primary difference is other countries track their students. Not just in separate classrooms, not a self-selected honours programme, but actually different high schools for people who score higher on the entrance exams.
236.
by programjames 1y ago
I saw an interesting post on LessWrong[1] that essentially argues education got so bad because all the education studies are using metrics biased towards the bottom. If you can't measure improvement in the top (10% for standardized tes
237.
by programjames 2y ago
Anyone willing to give an intuitive summary of what they did mathwise? The math in the paper is super ugly to churn through.
238.
by programjames 2y ago
The proper way to make lossy encoders is to minimize the mutual information between the original image and the encoding, while minimizing the necessary error correction between the reconstructed image and the original image. It's diffi
239.
by programjames 2y ago
I agree, z should be absorbed into μ and Σ, e.g. you always input `[1 0 0 ... 0]`, and the first layer of the neural network would essentially output z. They would have to stop approximating the KL(q(θ)||p(θ)) as O(θ^2) though, so maybe thi
240.
by programjames 2y ago
> Intelligence is compressing information into irreducible representation I thought that was physics ;) https://en.wikipedia.org/wiki/Wigner%27s_classification

More ›