Working version

test_parascore.py

@@ -119,16 +119,16 @@ class test_parascore(evaluate.Metric):
             elif div < thresh:
                 ss = -1 + ((thresh + 1) / thresh) * div
             diversity.append(ss)
-        return diversity[0]
+        return diversity
 
     def _compute(self, predictions, references, lang='en'):
         """Returns the scores"""
         
         score = self.score.compute(predictions=predictions, references=references, lang=lang)
-        bert_score = round(score['f1'], 2)
+        bert_score = [round(v, 2) for v in score['f1']]
         diversity = self._diverse(predictions, references, lang)
 
-        score = bert_score + 0.05 * diversity
+        score = [s + 0.05 * d for s, d in zip(bert_score, diversity)]
         return {
-            "accuracy": score,
+            "score": score,
         }