Function no_grad 

pub fn no_grad<F, R>(f: F) -> R
where
    F: FnOnce() -> R,

Execute a closure without gradient tracking.

Useful for inference or when gradients are not needed.

Example

use aprender::autograd::{Tensor, no_grad};

let x = Tensor::from_slice(&[1.0, 2.0]).requires_grad();

// No gradients computed inside this block
let y = no_grad(|| {
    x.mul(&x).sum()
});

assert!(y.grad().is_none());