nsor([[0, 1, 0], [1, 0, 1]])
        >>> preds = tensor([[0.11, 0.22, 0.84], [0.73, 0.33, 0.92]])
        >>> multilabel_matthews_corrcoef(preds, target, num_labels=3)
        tensor(0.3333)

    Nr7