Libraries for Python programming in GPU

I am working on converting Python code for data cleaning and pre-processing which were earlier running in CPU, to be run with Nvidia GPU.

I am using NumbaPro to write Python code for GPU. But, even simple tasks like finding average is really complex to be converted with help of Numba.

I see a lot of Cuda libraries to be used with C Programming (like cuBLAS), but not a lot with Numba or Python.

Is there any libraries that will accelerate GPU development with Python apart from Numba. I know Tensorflow for GPU helps in training models, but I am looking more of data pre-processing libraries like Python and Numpy for GPU.

This forum supports the use of the NVIDIA GPU Cloud. Please post your question on the CUDA Programming and Performance forum at https://devtalk.nvidia.com/default/board/57/cuda-programming-and-performance/