Visuals are amazing, as always!
I've recently learned that 3b1b has his own animation library.
https://github.com/3b1b/manim
I wonder if there are any manim art out there :)
> In contrast, running the neural network takes 5μs (twenty thousand times faster) with only a small loss in accuracy. This “approximate function inversion via gradients” trick is a very general one that can not only be used with dynamical systems, but also lies behind the fast style transfer algorithm.
Very interesting! Follow up question -- how would you choose the network architecture?
Since the network only acts on a small portion of the entire system, we can constrain it in such a way that dramatically simple NNs work just fine.
`FastChain(FastDense(3,32,tanh), FastDense(32,32,tanh), FastDense(32,2))` (from [0]) would take three inputs from your basis, run it through one hidden layer and provide you with two trained parameters.
This [1] example uses two hidden layers, its one of the more complex solutions I've seen so far. To move to this complexity from a simpler chain, we first make sure our solution is not in a local minima [2], then proceed to increase the parameter count if the NN fails to converge.
StyleGAN2 takes 512 dimensional vector as an input. Would be neat to attach some physical input device, like a matrix of dials... And encode your own painting with it :)
If there are no graphs or formulas in the book, I prefer to use a text-to-speech engine. Over the years I've become able to listen to it on great speeds. It feels easier to comprehend spoken language.
I listen to books and articles via text to speech application for over 3 years. Currently i am able to listen comfortably at 3.6x speed. I don't know how to test comprehension through. My friend says i might be able to think faster than other people, because of superhuman listening powers i now posess. Don't know how to test that either.
I have made a browser extension that helps me put every webpage I read through TTS. I also get current word highlight on text and automatic page scroll. It helps me focus twice as good, I can't even stand reading without TTS support any more.
Unfortunately I can only stand one voice - "Alex" from OSX, at about 1.3x ... 1.5x speed. Other voices I find irritating. I'm following the new WaveNet TTS voices with great expectations, but they don't seem to come near my laptop.
