Closed andreufont closed 1 year ago
When running the forecast code of cup1d with LaCE-NN, I get the following error:
RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
I thought we were not using CUDA when running with CPUs?
Laure found a way to avoid this by setting os.environ['CUDA_VISIBLE_DEVICES'] = ""
I have this now in my .bashrc: export CUDA_VISIBLE_DEVICES=""
When running the forecast code of cup1d with LaCE-NN, I get the following error:
RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
I thought we were not using CUDA when running with CPUs?