1

I am new to RAPIDS, and just tried to replicate a logisticregression model previously with sklearn, but now with cuml, but it the LogisticRegression() call does not work.

from cuml.linear_model import LogisticRegression, LinearRegression
ols_cuml=LogisticRegression(penalty='l2')

ols_cuml.fit(X,y)

I get the following error:

python-error with cuml

the LinearRegression() load statement also gets the same error. Not sure why this is happeding considering cudf(Pandas for CUDA) loads well, and can operate on gpu dataframes.

talonmies
  • 70,661
  • 34
  • 192
  • 269
alf
  • 11
  • 2

1 Answers1

0

The above error that you mentioned generally happens if the build or python setup runs into errors.

I am using cuml-0.15 (nightly release) and dont see the same error as the one you mentioned above. Try installing cuml from the command provided here : https://rapids.ai/start.html

The one that I used is : conda install -c rapidsai-nightly -c nvidia -c conda-forge \ -c defaults cuml=0.15 python=3.7 cudatoolkit=10.2

saloni
  • 296
  • 1
  • 7