ic1149
commented
2 weeks ago Don't use anaconda, don't use python 3.11 , use vanilla python 3.9 . Some libraries doesn't work with cython/ python 3.11, which anaconda is using. anaconda also has different file pathing, the base directory isn't at your python file like vanilla python, instead it is the anaconda folder.
I'm a total beginner to Python as a whole, and I'm using Python version 3.11 ( ran it in ANACONDA) the pytorch/torchvision/torchaudio part went well, until it was where i was supposed to install the dependencies using "requirements.txt" with pip. since i have a NVIDIA GPU, I used the one with "pip install -r requirements.txt" as the manual suggested. However, this error with the " sklearn" happened. i have no idea what happened at all, and i'm trying my best to resolve it. i have went to the https://numpy.org/devdocs/reference/distutils_status_migration.html site, but sadly i couldn't really understand anything as i'm not very experienced.
then something similar to " performance hint: sklearn_loss_loss.pyx:499:5: Exception check on 'cgrad_hess_half_binomial' will always require the GIL to be acquired. Declare 'cgrad_hess_half_binomial' as 'noexcept' if you control the definition and you're sure you don't want the function to raise exceptions." message kept appearing.
and this as well.
i skipped a few lines that basically displays the same message.
i can always provide the full message if it's useful.
any help would be greatly appreciated!!