Some explenation about the L2 norm:
The L2 norm reduces the dimension of a multi dimensional vector to 1, e.g. in physical space, which is a vector of shape 3, the L2 distance has shape 1 (which we use as the distance in every day live, if you measure the distance (shape 1), between two objects in your room, each with coordinates of shape 3).
You can also calculate this for other cases than the 3 dimensional, but the output will always be of shape 1, e.g. for a vector of shape 256, the output will be 1d. You could also do this individually, i.e. see your 256 entries as 256 1d vectors, then the output shape would be 256. In the 1d case the L2 norm is the same as the absolute value.
Maybe you could specify: Do you want to know how to calculate the L2 distance? Or are you looking for another measure?
For 1d coordinates the absolute value is the common L2 distance. If you see your second dimension as 256 seperate 1d coordinates, then your code computes you the 256 corresponding distances. But: It is not the L2 norm of a 256d vector.