nilslacroix

I found the reason for the error: When using multiple GPU devices with `devices = "0:1:2"` for example this seems to distribute loads when you look at` nvidia-smi`, but in...

I solved this issue myself. The problem occurs if for example in a scikit gridsearch` n_jobs = -1` is set, which is pretty much standard, to use all cpu cores....

Apparently and strangely enough the fitting was fastest with n_jobs=1 for a 1million samples*40 columns dataset. This does not occur with other GPU related algorithms in scikit (like xgboost for...

Actually I don't know yet how to reproduce the error. When I do not use the code line `env = ch.envs.ActionSpaceScaler(env)` the example runs perfectly fine. A collegue of mine...

Is this being adressed or do you know a workaround for the problem?

Thanks for the effort, this works for my target encoder, but it fails for One Hot Encoding or Binary Encoder. For example when i encode one column with `LeaveOneOutEncoder()` and...

Okay I just took a look at the code of category encoders and changed your class to: ``` from category_encoders import BinaryEncoder as BE class BinaryEncoder(BE): def get_feature_names_out(self, input_features=None): """Get...

I can provide this screenshot if this is good?  Also for the dtypes: ``` Nightlife float64 Education float64 Security float64 Nature float64 Eating float64 Balcony int64 DailyNeeds float64 Tourism...

[dataset.zip](https://github.com/scikit-learn/scikit-learn/files/8661133/dataset.zip) This is the minimum size where I could reproduce the error. Strangely enough it does not happen with smaller datasets, maybe this is linked to samplesize.

The problem seems to be within column "YearModernization". The warning does not appear when I use the power transformer outside of the pipeline: So this works: ``` pt = PowerTransformer(method="yeo-johnson")...