Nick Becker

As a note, cuML's TreeExplainer is built on top of GPUTreeShap and also [doesn't support categorical features today](https://docs.rapids.ai/api/cuml/nightly/api.html#cuml.explainer.TreeExplainer). ``` Parameters: model object The tree based machine learning model. XGBoost, LightGBM,...

This comes from partd. Solving this would also help enable `groupby.apply` with cudf backed dask dataframes: https://github.com/rapidsai/cudf/issues/5755#issuecomment-976823896

Nice! Are you saying the traceback you're now seeing still references the removed internal pandas functionality? Do we possibly need to update a dependency pinning somewhere?

The `percentile(col)` use case may not be blocked by https://github.com/rapidsai/cudf/issues/4706 , as the top-level DataFrame has a quantile implementation and delegates appropriately. ```python import dask import dask_cudf ​ ddf =...

Rather than use the RAPIDS Colab script, I recommend combining the suggestion above with [this comment about SageMaker Studio Lab](https://github.com/MaartenGr/BERTopic/issues/495#issuecomment-1150057704) (perhaps using new versions as needed/useful)

cuML is currently experimentally supported on Windows via Windows Subsystem for Linux (WSL). Installation instructions are linked at https://rapids.ai/start.html. We're (RAPIDS) actively working on improving and streamlining this WSL support,...

@cjnolet , was this resolved with the related #2027?

@dantegd is this still relevant tech debt?

Thanks for filing this feature request. This is a limitation of the current approach. Often when the data transfer time is non-trivial the estimator training time is the bulk of...

@divyegala CuPy arrays will fail the internal `_validate_data` checks in scikit-learn that ultimately call down to `np.asarray` and run into the equivalent implicit conversion error. There's been some recent discussion...