update app

app.py

@@ -178,15 +178,17 @@ def layer_sustainability_analysis(dataset_name, model_name, selected_attacks, nu
     logs = ["BSM:: experiment is being started ..."]
     device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
 
+    logs.append(f"Loading {dataset_name} dataset...")
     dataset, _ = get_dataset_and_transform(dataset_name)
     testloader = torch.utils.data.DataLoader(dataset, batch_size=32, shuffle=False)
-    logs.append(f"{dataset_name} dataset loaded")
+    logs.append(f"{dataset_name} dataset loaded with {len(testloader)} batches.")
 
+    logs.append(f"Initializing model {model_name} on {device}...")
     model = initialize_model(model_name, device)
-    logs.append(f"Model {model_name} loaded on {device}")
+    logs.append(f"Model {model_name} initialized.")
 
     param_count, layer_count = get_model_stats(model)
-    logs.append(f"Model stats: {param_count} parameters, {layer_count} layers")
+    logs.append(f"Model stats: Parameters = {param_count}, Layers = {layer_count}")
 
     all_attacks = {
         'FGSM': FGSM(model, eps=0.03),
@@ -204,21 +206,27 @@ def layer_sustainability_analysis(dataset_name, model_name, selected_attacks, nu
     timestamp = datetime.now().strftime('%Y%m%d_%H%M%S')
     output_dir = os.path.join(output_dir_base, f"{model_name}_{timestamp}")
     os.makedirs(output_dir, exist_ok=True)
-    logs.append(f"Output directory: {output_dir}")
+    logs.append(f"Output directory created: {output_dir}")
 
     results = {atk: {'cm': [], 'mvl': []} for atk in attacks}
 
     for i, (images, labels) in enumerate(testloader):
         if i >= num_batches:
+            logs.append(f"Reached batch limit: {num_batches}")
             break
         images, labels = images.to(device), labels.to(device)
         logs.append(f"Processing batch {i+1}/{num_batches}...")
 
         for atk_name, atk in attacks.items():
+            logs.append(f"  Running attack: {atk_name} on batch {i+1}")
             adv_images = atk(images, labels)
             mvl_vals = compute_mvl(model, images, adv_images, device)
             results[atk_name]['mvl'].append(mvl_vals)
-            results[atk_name]['cm'].append(np.mean(mvl_vals))
+            batch_cm = np.mean(mvl_vals)
+            results[atk_name]['cm'].append(batch_cm)
+            logs.append(f"    Attack {atk_name}: batch CM={batch_cm:.6f}")
+
+    logs.append("Finished processing batches, computing statistics...")
 
     cm_means = {atk: np.mean(results[atk]['cm']) for atk in attacks}
     cm_stds = {atk: np.std(results[atk]['cm']) for atk in attacks}
@@ -279,6 +287,7 @@ def layer_sustainability_analysis(dataset_name, model_name, selected_attacks, nu
     logs.append(f"Saved integrated MVL plot: {integrated_mvl_plot_path}")
 
     processing_time = time.time() - start_time
+    logs.append(f"Processing completed in {processing_time:.2f} seconds")
 
     stats = {
         'Dataset': dataset_name,
@@ -309,7 +318,7 @@ paper_info_html = """
 <div style="border: 1px solid #ccc; padding: 15px; border-radius: 8px; margin-bottom: 15px;">
   <h2>Layer-wise Regularized Adversarial Training Using Layers Sustainability Analysis Framework</h2>
   <h3>Authors</h3>
-  <p>Mohammad Khalooei, Mohammad Mehdi Homaypour, Maryam Amirmazlaghani</p>
+  <p>Mohammad Khalooei, Mohammad Mehdi Homayounpour, Maryam Amirmazlaghani</p>
 
   <h3>Abstract</h3>
   <ul>
@@ -348,7 +357,7 @@ def create_interface():
         model_text = gr.Textbox(value="LeNet", visible=False, interactive=False, label="Model")
 
         attack_input = gr.CheckboxGroup(choices=attacks, label="Select Attacks", value=attacks)
-        batch_input = gr.Slider(minimum=1, maximum=20, step=1, value=5, label="Number of Batches")
+        batch_input = gr.Slider(minimum=1, maximum=20, step=1, value=2, label="Number of Batches")
         run_button = gr.Button("Run Analysis")
 
         error_output = gr.Textbox(label="Error", visible=False)