This PR introduces the approximate computation of inverse-hessian-vector-products, useful for calculating influence vectors and values. By using forward-over-backward autodiff combined with a conjugate gradients solver, we are able to compute these quantities without the need for storing or inverting the hessian matrix.
This PR introduces the approximate computation of inverse-hessian-vector-products, useful for calculating influence vectors and values. By using forward-over-backward autodiff combined with a conjugate gradients solver, we are able to compute these quantities without the need for storing or inverting the hessian matrix.