PermutationImportance returns score, larger than 1

[lyriccoder]

I have trained a RandomForest model (CV_rfr2.best_estimator_). Then when I run PermutationImportance, I get the 1.01319077e+00 score of a feature :

perm2 = PermutationImportance(CV_rfr2.best_estimator_, cv = "prefit", refit = False, n_iter = 100).fit(X_train2, y_train2)
perm2.feature_importances_

array([5.82195156e-03, 1.22124533e-17, 1.33226763e-17, 2.22044605e-17,
       2.96321815e-04, 6.68249047e-04, 1.99840144e-17, 2.22044605e-17,
       8.13244234e-05, 2.55351296e-17, 8.54808844e-05, 9.99385230e-04,
       8.48752514e-03, 1.03550714e-01, 3.65605539e-04, 4.57918932e-03,
       6.08199650e-03, 8.17786603e-04, 7.77156117e-18, 9.79967255e-04,
       1.44328993e-17, 2.44249065e-17, 2.10942375e-17, 3.25723322e-03,
       1.01319077e+00, 1.49282346e-02, 5.51620039e-02, 3.69313730e-02])

Is it ok? It seems that the value shouldn't be more than 1.