Chewxy

More vectorized instructions were added to `vecf32` and `vecf64`. Will have a few more updates to those packages

Which functions do you have in mind? I don't mind changing the function signatures if it means better error reporting to the user. I do think that `NewNode()` and exported...

Agreed.

Experimental only. Culmination of rethinking that happened around #65

Yeah... I was doing it by hand. There are a number of circular dependencies (specifically the math ops - the symbolic differentiation bits relies quite heavily exposed math API, which...

I'm not too worried about the task queue. That's the easiest part of the equation - users should be free to choose between any communications protocol they desire - goroutine,...

Will investigate and fix once I fix my machine

FYI my home machine still hasn't has its graphivcs card fixed.

the values cannot be accessed because a threadlock is required to access CUDA context.

Seems trivial enough. The trivial solution is to just use masking (elemwise operations). This ensures that it's entirely differentiable. However if you have a link to the calculation of the...