covariance matrix of the multi-output GP regression using GPflow

Question

I just followed the guidance for multi-output GP regression

Here, I need to get the covariance across different dimensions of the output. For example, my output (denoted as Y) is 2D, and there are 100 sample points. Then I need the 100 2-times-2 covariance for Y.

How can I do it easily in the GPflow?

I've tried m.kernel(X), but it seems incorrect...