TimotheeMathieu

Yes sorry, the release was almost done (the binaries are artifacts of PR #156 ). I was busy at work and forgot the last step of uploading to pypi. I...

I did the upload. We have binaries for python 3.6 to python 3.9. Not yet for python 3.10. @0xb39 could you retry `pip install scikit-learn-extra` now to see if it...

I will try to build it.

After some work (cibuildwheel is not easy to debug) I managed to have python 3.10 and 3.11 for 64bit. I uploaded them on pypi. This should be sufficient for now...

For now you can try the method I suggest in issue #177, it is not ideal but at least on my computer it installs for python 3.12

Hi @aaaimee000, could you try the methods suggested in #177 ? In particular, try to download the wheel from the artifact in the action (https://github.com/scikit-learn-contrib/scikit-learn-extra/actions/runs/11686011469) and install directly from the...

Hello. Apparently, there is a deprecated cython call in there that is messing with the install. I think I fixed it in my fork (or at least I was able...

From the tests I did, it seems to be a conflict between python multiprocessing and pytorch multiprocessing. I just tried by replacing everything multiprocessing in AgentManager by joblib and there...

I understand the idea but these kwargs are very cryptic to newcommers (and even for me, I have to read the code every time to know what can go in...

My idea was the following: - Have an optional argument in the fit that default to None, as you said it can be something like `stop_callback: None or Callable[[Writer], bool]`....