Fix bug in evaluation metrics

RUC-NLPIR · Jul 31, 2024 · d0f403e · d0f403e
1 parent 785a901
commit d0f403e
Showing 1 changed file with 4 additions and 3 deletions.
diff --git a/flashrag/evaluator/metrics.py b/flashrag/evaluator/metrics.py
@@ -30,16 +30,17 @@ def calculate_metric(self, data):
         return {}, []
 
     def get_dataset_answer(self, data):
-        if data.choices != []:
+        if any(choice == [] for choice in data.choices):
+            golden_answers_list = data.golden_answers
+        else:
             # multi-choice dataset
             all_choices_list = data.choices
             golden_choice_idx_list = data.golden_answers
             golden_answers_list = [
                 [choices[idx] for idx in idx_list]
                 for choices, idx_list in zip(all_choices_list, golden_choice_idx_list)
             ]
-        else:
-            golden_answers_list = data.golden_answers
+
         return golden_answers_list