Fabrice Normandin

Hello! Thanks very much for posting this, and sorry I didn't reply earlier. I might slightly alter the formatting in the text above, if you don't mind. I seem to...

Hey @rggjan, thanks for posting. At the moment, if the default is True, then the action is "store_false" and vice-versa. It's kind-of up to the user to use a good...

> can you give me a single example where the current behavior makes sense for a default which is True? Good point! 1. For the default value, the main "culprit"...

Sorry this has taken a while. The PR for this (#94) was getting quite stale, and I didn't like having a custom BoolAction class for it. Ok I'll push this...

Hey there @nacitar , sorry for replying earlier. This is a cute, crunchy little bug. I really like those. For anyone that wants to help out, this is what I'm...

Good point! I think the codebase currently heavily depends on the annotations being evaluated (e.g. on the `type` attribute of the `dataclasess.Field` objects to be a 'live' type (`` and...

Hey @Conchylicultor , No updates yet, I'm quite busy with the ICLR deadline at the moment. I do plan to get to this at some point, but I could also...

I have a solution in mind: Setting `max_trials` on the SpaceTransform wrapper should only set it on the wrapper itself. It shouldn't be propagated down onto the wrapped algorithm.

We currently set that value with `algo.unwrapped.max_trials`, it would instead be `algo.max_trials = max_trials`, and the AlgoWrapper class would have a max_trial property with a custom setter that (by default)...

Was Orion installed in editable mode, in the environment where this is run? This would explain the "no module named orion.algo.hebo" error. Otherwise, I'm not sure I understand what's going...