Optimizers

cargo add zyx-optim

Optimizers take gradients calculated as your loss w.r.t. your parameters and update those parameters so that the next time you run your model with the same inputs, the loss will be lower.

#![allow(unused)]
fn main() {
let mut optim = zyx_optim::SGD { ..Default::default() };
let grads = loss.backward();
optim.update(&mut model, grads);
}

Zyx has multiple optimizers. All are accessible from crate zyx-optim.