Update app.py

app.py

@@ -641,8 +641,40 @@ with gr.Blocks(css=css_tech_theme) as demo:
                 eval_status = gr.Textbox(label="🛠️ Evaluation Status", interactive=False,scale=1,min_width=1200)
         
         # Define the functions outside the `with` block
+        # def handle_evaluation(file, model_name):
+        #     # Check if required inputs are provided
+        #     if not file:
+        #         return "Error: Please upload a prediction file.", 0, gr.update(visible=False)
+        #     if not model_name or model_name.strip() == "":
+        #         return "Error: Please enter a model name.", 0, gr.update(visible=False)
+        
+        #     try:
+        #         # Load predictions file
+        #         predictions_df = pd.read_csv(file.name)
+        
+        #         # Validate required columns in the prediction file
+        #         required_columns = ['question_id', 'predicted_answer']
+        #         missing_columns = [col for col in required_columns if col not in predictions_df.columns]
+        #         if missing_columns:
+        #             return (f"Error: Missing required columns in prediction file: {', '.join(missing_columns)}.",
+        #                     0, gr.update(visible=False))
+        
+        #         # Perform evaluation
+        #         status, leaderboard = evaluate_predictions(file, model_name, add_to_leaderboard=False)
+        #         if leaderboard.empty:
+        #             overall_accuracy = 0
+        #         else:
+        #             overall_accuracy = leaderboard.iloc[-1]["Overall Accuracy"]
+        
+        #         # Show the submit button after successful evaluation
+        #         return status, overall_accuracy, gr.update(visible=True)
+        
+        #     except Exception as e:
+        #         # Handle unexpected errors
+        #         return f"Error during evaluation: {str(e)}", 0, gr.update(visible=False)
+
         def handle_evaluation(file, model_name):
-            # Check if required inputs are provided
+    # Check if required inputs are provided
             if not file:
                 return "Error: Please upload a prediction file.", 0, gr.update(visible=False)
             if not model_name or model_name.strip() == "":
@@ -660,18 +692,16 @@ with gr.Blocks(css=css_tech_theme) as demo:
                             0, gr.update(visible=False))
         
                 # Perform evaluation
-                status, leaderboard = evaluate_predictions(file, model_name, add_to_leaderboard=False)
-                if leaderboard.empty:
-                    overall_accuracy = 0
-                else:
-                    overall_accuracy = leaderboard.iloc[-1]["Overall Accuracy"]
-        
+                results = evaluate_predictions(file, model_name, add_to_leaderboard=False)
+                overall_accuracy = results['overall_accuracy'] * 100  # Convert to percentage
+                
                 # Show the submit button after successful evaluation
-                return status, overall_accuracy, gr.update(visible=True)
+                return "Evaluation completed.", overall_accuracy, gr.update(visible=True)
         
             except Exception as e:
                 # Handle unexpected errors
                 return f"Error during evaluation: {str(e)}", 0, gr.update(visible=False)
+
         
         def handle_submission(file, model_name):
             # Handle leaderboard submission