Meraj Hashemizadeh

I will add new test cases for multi-dimensional outputs🙂

## `test_parafac` ``` _______________________ test_parafac[svd-3-5-True-True] ________________________ linesearch = True, orthogonalise = True, true_rank = 3, rank = 5, init = 'svd' monkeypatch = @pytest.mark.parametrize("linesearch", [True, False]) @pytest.mark.parametrize("orthogonalise", [True, False]) @pytest.mark.parametrize("true_rank,rank",...

In Numpy `float` represents double numbers. When using Pytorch a floating-point overflow happens somewhere in the calculations. To overcome this just replace `t = tl.tensor(t, device='cuda:0')` with `t = tl.tensor(t,...

As the skipped dimensions are included in the shape of the original tensor, there is no need for the `skip_end` argument to be passed to the function. @JeanKossaifi, should we...