Update compute_score.py

compute_score.py

@@ -84,9 +84,11 @@ def compute_score(dataset, predictions):
                 ground_truths = list(map(lambda x: x["text"], qa["answers"]))
                 prediction = predictions[qa["id"]]
                 exact_match += metric_max_over_ground_truths(exact_match_score, prediction, ground_truths)
-                precision += metric_max_over_ground_truths(precision_score, prediction, ground_truths)
-                recall += metric_max_over_ground_truths(recall_score, prediction, ground_truths)
-                f1 += (2 * precision * recall) / (precision + recall)
+                precision_qa = metric_max_over_ground_truths(precision_score, prediction, ground_truths)
+                recall_qa = metric_max_over_ground_truths(recall_score, prediction, ground_truths)
+                f1 += (2 * precision_qa * recall_qa) / (precision_qa + recall_qa)
+                recall += recall_qa
+                precision += precision_qa
     exact_match = 100.0 * exact_match / total
     f1 = 100.0 * f1 / total
     recall = 100.0 * recall / total