Function autograd::ops::grad_with_default [−] [src]

pub fn grad_with_default(
    ys: &[&Tensor], 
    xs: &[&Tensor], 
    output_grads: &[&Tensor]
) -> Vec<Tensor>

Returns gradient tensors wrt input tensors.

Arguments

  • ys - Targets of differentiation.
  • xs - tensors with which differentiate ys.
  • output_grads - Already known gradients of ys.

The length must be same as ys's. If each objective is not a scalar, you must pass the "Some" value. In most cases, it is initialized with 1s.

Returns

Symbolic gradient tensors corresponding to xs in the same order as xs

For detailed, see grad.