Malte Londschien

Hi Harry Is this in Python or R? In Python, you can access the p-value via the `p_value` attribute: ```python In [1]: import numpy as np ...: ...: Sigma =...

Thanks Jan! ``` mlondsch@nid005009:~/code/$ getconf PAGESIZE 65536 ``` A quick ``` export JEMALLOC_SYS_WITH_LG_PAGE=16 pip install tabmat-4.0.1.tar.gz --no-deps ``` raised an error, I'll have a closer look tomorrow. ``` mlondsch@nid005009:~/code/$ pip...

Same as before. ```python (base) mlondsch@nid005008:~/code$ conda install -c conda-forge tabmat=4.0.1=py312ha2895bd_1 Channels: - conda-forge Platform: linux-aarch64 Collecting package metadata (repodata.json): done Solving environment: done ## Package Plan ## environment location:...

Following the installation steps of `tabmat` in https://github.com/Quantco/tabmat/blob/e288b47bfbf04c92ce824008f0cf615d32bcbad7/pyproject.toml#L81-L95 (as suggested by the awesome cscs support) ``` git clone --branch 5.3.0 https://github.com/jemalloc/jemalloc.git cd jemalloc ./autogen.sh --disable-cxx --with-jemalloc-prefix=local --with-install-suffix=local --disable-initial-exec-tls make make...

No. It's also not set by default. ``` mlondsch@nid005009:~/code$ echo $JEMALLOC_SYS_WITH_LG_PAGE ``` edit: below are the full build logs, in case this helps ``` mlondsch@nid005009:~/code$ git clone --branch 5.3.0 https://github.com/jemalloc/jemalloc.git...

> This actually looks pretty good, it's just that on the inference side, we're still expecting doubles in a lot of places. I think there are also quite some "Kinderkrankheiten"...

Two questions about the convergence criteria: - I know that even if `scale_predictors = True`, internally, data and coefficients are kept without rescaling, but the lasso / ridge penalties are...

> I think it would make sense to extend the test suite to float32s before merging this (see [653d6f1](https://github.com/Quantco/glum/commit/653d6f1b00f436f0d175355bc89124ed21210a1c)). There are still a few places where we are mixing dtypes,...

@MarcAntoineSchmidtQC can we merge this?

See https://github.com/Quantco/glum/issues/872 for an example for "bad convergence" with `float64`. I guess this would only be worse for `float32`. I'm not sure how to turn this into a test.