Retrieve a neural network layer from the
mlr_pipeops dictionary.
The id of the returned PipeOp is .key.
Because the ids within a Graph must be unique, .key can be suffixed
with _<n> to disambiguate repeated layers, e.g. nn("linear_1") and nn("linear_2").
Examples
po1 = po("nn_linear", id = "linear")
# is the same as:
po2 = nn("linear")
# the `_<n>` suffix is part of the id, but not of the dictionary key
nn("linear_1")$id
#> [1] "linear_1"