Pytorch: computing the eigenvector of data point

Question

Lets say we already have estimated the principal directions and eigenvalues of a dataset X using pytorch:

U, S, V = torch.svd(X)