Have you heard of Bend? I didnt before yesterday evening. That is massively parallel, high-level programming language, that just a set of Python style bindings for Rust parallelization bindings.
Have you heard micrograd? That is, citing A tiny scalar-valued autograd engine and a neural net library on top of it with PyTorch-like API. Sounds cool!
What sound even cooler? micrograd written in two day old language!
Im not sure I wrote it idiomatically and I cant really benchmark it, bc it doesnt even has IO, but Im looking forward to it.
Besides that, it doesnt has random generator, so I wrote some classic pseudorandom.
So, how do I use it? Other than installing Rust and Bend, clone this repo and add some neurons
import "nn.bend" exposing (MLP, Value)
def main():
# Example usage
mlp = MLP.new(3, [4, 4, 1])
input_values = [Value.new(1.0), Value.new(2.0), Value.new(3.0)]
output = mlp.call(input_values)
print(output) # 🙏🙏🙏
main()