Update evaluation.py

evaluation.py

@@ -96,18 +96,24 @@ def compute_rmse(predicted_values, ground_truth_values):
     else:
         print("Invalid input for RMSE calculation. Ensure all values are numeric.")
         return None
-    
+def convert_adherence_to_numerical(adherence_score):
+    if adherence_score:
+        return 1  # True becomes 1
+    else:
+        return 0  # False becomes 0
+        
 def retrieve_ground_truths(question, dataset):
     """Retrieve the ground truth answer for a given question from the dataset."""
     for split_name, instances in dataset.items():
         for instance in instances:
             if instance['question'] == question:
                 instance_response = instance['response']
+                adherence_numerical = convert_adherence_to_numerical(instance['adherence_score'])
                 ground_truth_metrics = {
                     "context_relevance": instance['relevance_score'],
                     "context_utilization": instance['utilization_score'],
                     "completeness": instance['completeness_score'],
-                    "adherence": instance['adherence_score']
+                    "adherence": adherence_numerical
                 }
                 return instance_response, ground_truth_metrics  # Return the ground truth response immediately
     return None,None  # Return None if no match is found