Spaces:
Running
Running
How to set CUDA_HOME when compile CUDA operations?
#160
by
allencbzhang
- opened
In my space, I need to set a variable to the path of CUDA, and then compile the repository. I try to use
export CUDA_HOME=$(dirname $(dirname $(which nvcc))) && pip install .
However, it doesn't work. Is there any way to set the CUDA_HOME?