torch.autograd.gradcheck.get_analytical_jacobian

torch.autograd.gradcheck.get_analytical_jacobian(inputs, output, nondet_tol=0.0, grad_out=1.0)