Fix prediction str -> list_of_list

app.py

@@ -734,6 +734,8 @@ with gr.Blocks(theme='shivi/calm_seafoam', css_paths='style.css', js=js_func) as
                 metrics_conc = target_df    
                 if 'valid_efficency_score' not in metrics_conc.columns:
                     metrics_conc['valid_efficency_score'] = metrics_conc['VES']
+                if 'VES' not in metrics_conc.columns:
+                    metrics_conc['VES'] = metrics_conc['valid_efficency_score']
                 eval_text = generate_eval_text("End evaluation")
                 yield gr.Markdown(eval_text, visible=True), gr.Image(), gr.Markdown(), gr.Markdown(), gr.Markdown(), metrics_conc, *[predictions_dict[model][columns_to_visulize] for model in model_list]
             
@@ -868,14 +870,15 @@ with gr.Blocks(theme='shivi/calm_seafoam', css_paths='style.css', js=js_func) as
                             db_path_name="db_path"
                         )
                     else: 
+                        
                         metrics_df_model = us.evaluate_answer(predictions_dict[model])
                     metrics_df_model['model'] = model
                     metrics_conc = pd.concat([metrics_conc, metrics_df_model], ignore_index=True)
 
-                if 'valid_efficency_score' not in metrics_conc.columns and 'VES' in metrics_conc.columns:
+                if 'valid_efficency_score' not in metrics_conc.columns:
                     metrics_conc['valid_efficency_score'] = metrics_conc['VES']
 
-                if 'VES' not in metrics_conc.columns and 'valid_efficency_score' in metrics_conc.columns:
+                if 'VES' not in metrics_conc.columns:
                     metrics_conc['VES'] = metrics_conc['valid_efficency_score']
                 
                 eval_text = generate_eval_text("End evaluation")

utilities.py

@@ -185,9 +185,10 @@ def evaluate_answer(df):
             target = row["target_answer"]
             predicted = row["predicted_answer"]
             try:
+                predicted = eval(predicted)
                 result = metric.run_metric(target = target, prediction = predicted)
             except Exception as e:
-                result = None
+                result = 0
             results.append(result)
         df[metric_name] = results
     return df