crf.py 文件源码

def generate_folds(labels_fname, folds_fname, max_n_folds=10):
    """
    Generate folds for CV exps with n = 2, ..., max_n_folds.
    Save as pickled dict with n as key.
    """
    filenames = read_labels(labels_fname)['__filenames__']
    folds = {}

    for n in range(2, max_n_folds + 1):
        # Create folds from complete texts only
        # (i.e. instances/sentences of the same text are never in different folds).
        # There is no random seed, because the partitioning algorithm is deterministic.
        group_k_fold = GroupKFold(n_splits=n)
        # Don't bother to pass real X and Y, because they are not really used.
        folds[n] = list(group_k_fold.split(filenames, filenames, filenames))

    print('writing folds to ' + folds_fname)
    pickle.dump(folds, open(folds_fname, 'wb'))