mb706
mb706
this is what currently breaks `as.data.table(mlr_pipelines)` I believe. Maybe use expressions or something.
Eventually we want to do smart caching like python Dask.
PipeOp that just prints info messages about the task (or anything) passing through. One input, one output. Printer settings should be set during construction, not through parameter set, since tuning...
PipeOps may have "optional" channels that are not used by the `%>>%` operator but can still be used by using `add_edge()`. This is useful to provide an additional result of...
The user gives a function as parameter that takes a task and returns a Graph. That graph is then trained on the training data and the trained Graph is used...
like PipeOpColApply but for rows (of a given type).
Maybe we should have a built-in task or two where the row-ids are not just `1:task$nrow`, so users don't get used to that assumption.
Loading e.g. a `Task` that was generated with an old mlr3-version using `readRDS()` will give errors like this: https://github.com/mb706/autoxgboost3/issues/2#issuecomment-596969950 . Maybe there is a way to make these kinds of...