BUG: scikit learn's metric f1 score now returns a python native float

BUG: scikit learn's metric f1 score now returns a python native float. Converting it back to a numpy array for backward compatibility

f1.py

@@ -14,6 +14,7 @@
 """F1 metric."""
 
 import datasets
+import numpy as np
 from sklearn.metrics import f1_score
 
 import evaluate
@@ -127,4 +128,5 @@ class F1(evaluate.Metric):
         score = f1_score(
             references, predictions, labels=labels, pos_label=pos_label, average=average, sample_weight=sample_weight
         )
+        score=np.array(score)
         return {"f1": float(score) if score.size == 1 else score}