RuntimeError: Cannot re-initialize CUDA in forked subprocess.

[MunirAbobaker]

CUDA Error inside multiprocessing.pool
I have the following method
x_ encoded = model.encode(x)
it is defined inside a method that is called using multiprocessing.pool and I get the following error from the encode method
RuntimeError: Cannot re-initialize CUDA in forked subprocess. To use CUDA with multiprocessing, you must use the 'spawn' start method
I used the spawn, but not solved

[shuoyes]

I have the same problem with you,could you tell me the details how to solve?