Thomas Moreau

You can do this with `-k`, as the extra parameters passed to `benchopt test` are passed to `pytest`

To deal with the `Nan`, we usually add a small `eps` to the curce so that it stays positive.

The command you are looking for is `benchopt plot --all .` I think. The goal of `generate-results` is to create a dashboard for the website. That being said, we should...

Ok but don't `benchopt plot --all .` generate the lasso webpage?

I think generating a plotting file is easy and could potentially convince many users to use `benchopt` for their papers' benchmark so I am +1 on this feature. If we...

Also note that we should not copy `static` in the output anymore. We should only take the static file from `ROOT/static` and embed them in the HTML.

Is this still an issue? The first appear to be due to the mismatch between version of editable install. this should have been fixed by #197. For the second one,...

A solution would be to add a `id` attribute to the `objective.py::Objective` that would serve as the name of the benchmark. else to avoi duplicating this, we could also use...

I think we cannot store it as a property of the dataset class because the optimal value depends on the `parameter/dataset`. What we could do is add an attribute to...

Also, I think this question is different from what @pierreablin was mentionning. For some benchmark, we might look at a scoring function that is in itself interesting (no need to...