glum with ray

[cfkstat]

Using glum and joblib with ray, I ran multiple models and found that threads could use 1 core, and if I set n_jobs=1, I could only use 50% of all the cpu。

[MarcAntoineSchmidtQC]

Can you share more of your setup and infrastructure? Maybe a minimal working example would also help to see how ray and glum interact. Unfortunately, I never used ray before, so I'm not sure what the intended behavior is.

[cfkstat]

When using ray, the value of the environment variable OMP_NUM_THREADS is set to 1 by default, and I changed it to 10, which worked fine in parallel. Is there any parameter in glum's package that can control the number of parallel cpus? If so, please tell me.

[MarcAntoineSchmidtQC]

OMP_NUM_THREADS is not specific to ray. It controls the number of cores used by openmp, which both ray and glum use. Setting the environment variable to 10 will provide 10 cores to glum. If you use the cross-validated version of GeneralizedLinearRegressor, you can also set n_jobs=X so that the cross-validation will be split across X cores.