I did not look at the source code yet, but is this some sorta physics modeling problem? Treat the strings as a a bunch of interconnected balls, then apply the force at point, compute other forces, interpolate a curve through the new state, then compute frequencies based on where the string was plucked? this is pretty cool

My guess is the strings are not modeled as balls balls as continuous lines with a set tension oscillating based on their length and thickness. It's presumably a bunch of trigonometry, but not the most insane modeling problem I've seen. Very cool none the less

What's the most insane one ?

The piano is a hard one. Best current result: Pianoteq.

https://www.modartt.com/pianoteq_overview

The question mark top left kind explains

> To me, LLM query management and retrieval is much more valuable than response generation

Sure, for certain tasks. For other tasks retrieval is less useful.

I think a superset of bash with strong types makes more sense than a brand new language that compiles to it. This is what makes typescript so successful (aside from other things) is that you can just drop it into your project and incrementally adopt it.

Call it "typebash" - easily understood by everyone

A bit of a clickbaity title, but I'll bite

I think "cleverness" is a function of your (and your team's) experience in a particular language / domain space

I worked with teams where verbose Java is considered "simple, straightforward, and easy to understand"

I also worked with teams where terse Haskell is considered "simple, straightforward, and easy to understand"

and of course these codebases look nothing alike

agreed, I find it pretty useful to check if I remember where LSTM stuff connects

I remember reading the original paper a while ago but always forget (pun intended) where to connect stuff

then I realized that memorizing it visually is not the best approach, it's better to think about it in this sorta loose fashion -- I remember there is forget gate, well it forgets previous stuff so there is probably some hadamard product somewhere, it probably needs some inputs and previous hidden states...there was some -1,1 forcing in candidate memory so probably needs tanh instead of sigmoid...and then piece by piece i can reconstruct it pretty closely

I didn't encounter the issue you described with adding an already existing edge though -- the counter stays the same

the answer to RNN is non-trivial, but perhaps Recurrent in RNN might be of some use

Even knowing what the R in RNN stands for requires some pre-existing knowledge of neural networks. Which isn't something that's helping _me_ learn about them, particularly.

Amazing

Now if only the education system improved as well and stopped feeding the kids useless information.

and both are backed by msft?

I never get tired of reading articles about typestate pattern

Valve GoldSrc for me