btschroer

You might want to cross-post this on the `pyarrow` repository. The community is very active and helpful. They might also be interested in this.

Can confirm, the following used to work in 1.5: ```python import polars as pl import pandas as pd t0 = pd.Timestamp("2001-01-03") frame = pl.from_dict({ "t": ["2001-01-01", "2001-01-02", "2001-01-03", "2001-01-04"] }).with_columns(...

👍 Yes, that would be really nice to have. I have a hierarchical model with lots of groups. @aseyboldt , @fonnesbeck: Is it at all possible that my `nutpie` sampling...

Thanks @aseyboldt for the suggestion, that's really helpful. I also looked into `compile_pymc_model` but it was essentially just calling functions "compile to numba" or "compile to jax", I am assuming...

`catboost` has added support for `numpy-2.0`, cf [here](https://github.com/catboost/catboost/issues/2671) and `ngboost` also supports `numpy-2.0` (https://github.com/stanfordmlgroup/ngboost/issues/358) now. Would be great to finally see support for `numpy-2.0` in `shap` now. Thanks a lot...