I am working with a > 5GB CSV file for competition in Kaggle. I am using cudf and cuml for data preprocessing and machine learning. but for visualization, my plan was to use GPU accelerated visualization using Plotly. Since Kaggle docker doesn't include Rapids yet, I am using the following Dataset to install it.
https://www.kaggle.com/cdeotte/rapids
But i am unable to import "cuxfilter" package in the Kaggle notebook even after installing rapids using the above dataset.
import cuxfilter
is giving ModuleNotFoundError
Is there any way to install this package in Kaggle?
I have also tried to install the package using conda but it was unsuccessful. The command used was the following
conda install -c rapidsai -c nvidia -c conda-forge -c defaults cuxfilter=0.15 python=3.7 cudatoolkit=10.0
Please also find my post in kaggle discussion forum form more details https://www.kaggle.com/cdeotte/rapids/discussion/190432#1046557
RAPIDS 0.15 release does not support CUDA10.0. Please update your CUDA package or use RAPIDS 0.14 release
Command to install cuxfilter 0.14 using conda :