This is a follow-up for mlpack/mlpack#2647. Based on the experiments in mlpack/mlpack#2777 we figured it makes sense to remove the boost::variant approach. This should not only make it easier for new-comers to contribute to the network codebase (the boost::variant and visitor approach can be confusing) but also reduce the memory usage during the bu

https://igel.readthedocs.io/en/latest/_sources/readme.rst.txt includes a link to the assets/igel-help.gif, but that path is broken on readthedocs.

readme.rst is included as ../readme.rst in the sphinx build.
The gifs are in asses/igel-help.gif

The sphinx build needs to point to the asset directory, absolutely:

.. image:: /assets/igel-help.gif

I haven't made a patch, because I haven't

I'm sorry if I missed this functionality, but CLI version hasn't it for sure (I saw the related code only in generate_code_examples.py). I guess it will be very useful to eliminate copy-paste phase, especially for large models.

Of course, piping is a solution, but not for development in Jupyter Notebook, for example.

String representations of dataset objects are used for previewing their contents from the terminal. When converting a Dataset object to a string, we build a table using ascii characters. The current table has fixed width columns that do not take full advantage of the terminal real estate if the dataset only contains a few columns.

echo $dataset;

<img width="574" alt="Annotation