Dear sklearn's experts,

Standard use of nested cross-validation within sklearn doesn't allow multi-core computing. As in the example below, njobs has to be set to 1 for inner/outer loops:

gs = GridSearchCV(pipe_svc, param_grid, scoring=score_type, cv, n_jobs=1)
scores = cross_val_score(gs, X, y, scoring, cv, n_jobs=1)

Would there be any no too difficult way to parallelize jobs in nested cross-validation, which would allow to highly reduce time-consuming computing ?

Thanks in advance !

Best,
Matthieu