Badr MOUFAD

Could you pls provide a link to a benchmark that reproduces the behavior?

This bug has more to do with the way exceptions are handled during solvers' runs. Indeed, we allow benchmarks to carry on despite errors (by showing `solver-name: error`). and that...

Here is [the preview](https://badr-moufad.github.io/share-benchmarks/preview_tags/benchmark_lasso_benchopt_run_2023-07-10_14h09m55.html) of the visualization feature, and the [benchmark repository](https://github.com/Badr-MOUFAD/benchmark_lasso/pull/4) to reproduce it. Feedback is very welcome, @svaiter. Also, how should we handle visualization conflicts between tags and...

Thanks for flagging this! I doubt that typed parameters would impair detecting/installing requirements. Could you pls provide a link to the CI error? Also, a link to the repository would...

Thank you again for your interest! Here are the key improvement factors - Levaraging the sparse nature of the solution with state-of-the-art working set strategy detailed in [our Neurips 2022...

If I got your point, you suggest shipping ``skglm`` with a pre-compiled ``numba`` code, right? It's a really tempting suggestion as it will eliminate the overhead of the first run....

For unittest, we need to add support for fix-point distance strategy to build working sets in the ``GroupBCD`` solver (later, add it to ``GroupProxNewton``) I opened an issue #208

Can you provide a case that reproduces this behavior? After fixing the bug, we shall use it to make a unitest.

To my knowledge, it would not be possible to share parameters in a URL because we lack a server that handles requests and determines how to serve the benchmarks to...

Right @mathurinm! It is just because we have a dedicated [parameter `references`](https://benchopt.github.io/how.html?highlight=reference) to put academic references? Should we drop it in favor of docstrings to provide descriptions?