Daniel Nagel
Daniel Nagel
I've needed to increase the `rtol` by a factor of 10 to make the tests pass for complex numbers. I have no idea if this is an issue of the...
I have also specified in the docs that the first two arguments of `np.linspace`, `np.logspace`, and `np.geomspace` have to be scalars. The numpy implementation supports also interpolation between arrays.
I need to admit that I was using geomspace only for real values. @guilhermeleobas thank you for finding this bug in my implementation and providing this helpful gist, but I...
> How exactly would the proposed feature be used? I admit I don't envision it clearly. What would the "different TOC" be? Would every module's TOC link to the extra...
One hack is to make the submodule itself private. Having no public classes or functions, this seems reasonable. But as you pointed out, this is only a hack and not...
I guess this is the relevant replacement function `sysconfig.get_path('platlib')`, see https://bugs.python.org/issue41282#msg393018 for further infromations.
Thank you for your kind answer and as you noticed correctly it is neither a security issue nor urgent.
I can recommend to simply apply the change, suggested in #731, manually. Changing a single line, solves all issues.
@seberg I was not aware that there is an character for null. `\0` is indeed a smart solution to my issue, thx a lot :)
Python 3.12 support has been added to scikit-learn, see scikit-learn/scikit-learn#27027. So this would be the first step towards supporting py3.12 with the latest sklearn 1.3