Another option for creating arrays Browser

[santi-mir]

Here you create 2d tensors but it is not really necessary or wise imho.

It would be nice to know why you preferred rank-2 tensors there.

If you have Rank 1 tensors (vectors) use cons xs = tf.tensor1d(array) or in this case a very handy function is const xs = tf.range(0,n) for creating dummy data.