Update app.py

app.py

@@ -139,14 +139,16 @@ def run_unlearning(index_to_unlearn, learning_rate):
     
     # Prediction before unlearning
     probs_before, pred_before = predict(original_model, image_tensor)
-    conf_before = probs_before[label_idx].item()
+    conf_actual_before = probs_before[label_idx].item()
+    conf_pred_before = probs_before[pred_before].item()
     
     # Perform the unlearning process on the duplicated model
     unlearn(unlearned_model, image_tensor, label_idx, learning_rate)
     
     # Prediction after unlearning
     probs_after, pred_after = predict(unlearned_model, image_tensor)
-    conf_after = probs_after[label_idx].item()
+    conf_actual_after = probs_after[label_idx].item()
+    conf_pred_after = probs_after[pred_after].item()
     
     print("Post-unlearning probabilities:", probs_after)
     
@@ -157,13 +159,17 @@ def run_unlearning(index_to_unlearn, learning_rate):
     result = f"""
 📍 Index Unlearned: {index_to_unlearn}
 🗂️ Actual Label: {label_name} (Index: {label_idx})
+
 🔎 BEFORE Unlearning:
-- Prediction: {cifar10_classes[pred_before]}
-- Confidence: {conf_before:.10f}
+- Predicted Class: {cifar10_classes[pred_before]} with confidence: {conf_pred_before:.10f}
+- Actual Class: {label_name} with confidence: {conf_actual_before:.10f}
+
 🧽 AFTER Unlearning:
-- Prediction: {cifar10_classes[pred_after]}
-- Confidence: {conf_after:.10f}
-📉 Confidence Drop: {conf_before - conf_after:.6f}
+- Predicted Class: {cifar10_classes[pred_after]} with confidence: {conf_pred_after:.10f}
+- Actual Class: {label_name} with confidence: {conf_actual_after:.10f}
+
+📉 Confidence Drop (Actual Class): {conf_actual_before - conf_actual_after:.6f}
+
 🧪 Test Set Performance:
 - Original Model: {orig_acc:.2f}% accuracy, Loss: {orig_loss:.4f}
 - Unlearned Model: {unlearn_acc:.2f}% accuracy, Loss: {unlearn_loss:.4f}