Trying to train using more than 1 parallel job on Windows results in the following error:
ImportError: [joblib] Attempting to do parallel computing without protecting your import on a system that does not support forking. To use parallel-computing in a script, you must protect your main loop using "if __name__ == '__main__'".
This is a known issue on Windows, and from the look of things will require a minor rework to wrap the training code in an "if __name__ == '__main__'"conditional. I'll work on it as I get free time, but if anyone else has a ready and working solution, it would make my life a bit easier.
In the meantime, I've changed
RandomizedSearchCV(classifier, parameters_to_try, n_jobs=globalArgs['numCPUs'], error_score=0, n_iter=n_iter, refit=True, cv=cvRounds)
to
RandomizedSearchCV(classifier, parameters_to_try, n_jobs=1, error_score=0, n_iter=n_iter, refit=True, cv=cvRounds)
and
searchCV = GridSearchCV(classifier, parameters_to_try, n_jobs=globalArgs['numCPUs'], error_score=0, refit=True, cv=cvRounds)
to
searchCV = GridSearchCV(classifier, parameters_to_try, n_jobs=1, error_score=0, refit=True, cv=cvRounds)
Trying to train using more than 1 parallel job on Windows results in the following error:
This is a known issue on Windows, and from the look of things will require a minor rework to wrap the training code in an
"if __name__ == '__main__'"conditional. I'll work on it as I get free time, but if anyone else has a ready and working solution, it would make my life a bit easier.In the meantime, I've changed
RandomizedSearchCV(classifier, parameters_to_try, n_jobs=globalArgs['numCPUs'], error_score=0, n_iter=n_iter, refit=True, cv=cvRounds)toRandomizedSearchCV(classifier, parameters_to_try, n_jobs=1, error_score=0, n_iter=n_iter, refit=True, cv=cvRounds)andsearchCV = GridSearchCV(classifier, parameters_to_try, n_jobs=globalArgs['numCPUs'], error_score=0, refit=True, cv=cvRounds)tosearchCV = GridSearchCV(classifier, parameters_to_try, n_jobs=1, error_score=0, refit=True, cv=cvRounds)