Update app.py

app.py

@@ -703,20 +703,20 @@ with gr.Blocks(css=css_tech_theme) as demo:
                 except Exception as e:
                     return f"Error loading ground truth: {e}", 0, gr.update(visible=False)
 
-        # Perform evaluation calculations
-        merged_df = pd.merge(predictions_df, ground_truth_df, on='question_id', how='inner')
-        merged_df['pred_answer'] = merged_df['predicted_answer'].apply(clean_answer)
-
-        valid_predictions = merged_df.dropna(subset=['pred_answer'])
-        correct_predictions = (valid_predictions['pred_answer'] == valid_predictions['Answer']).sum()
-        total_predictions = len(merged_df)
-
-        overall_accuracy = (correct_predictions / total_predictions * 100) if total_predictions > 0 else 0
-
-        return "Evaluation completed successfully.", overall_accuracy, gr.update(visible=True)
-
-    except Exception as e:
-        return f"Error during evaluation: {str(e)}", 0, gr.update(visible=False)
+                # Perform evaluation calculations
+                merged_df = pd.merge(predictions_df, ground_truth_df, on='question_id', how='inner')
+                merged_df['pred_answer'] = merged_df['predicted_answer'].apply(clean_answer)
+        
+                valid_predictions = merged_df.dropna(subset=['pred_answer'])
+                correct_predictions = (valid_predictions['pred_answer'] == valid_predictions['Answer']).sum()
+                total_predictions = len(merged_df)
+        
+                overall_accuracy = (correct_predictions / total_predictions * 100) if total_predictions > 0 else 0
+        
+                return "Evaluation completed successfully.", overall_accuracy, gr.update(visible=True)
+        
+            except Exception as e:
+                return f"Error during evaluation: {str(e)}", 0, gr.update(visible=False)