diff --git a/01.train_ODC_CNN.ipynb b/01.train_ODC_CNN.ipynb
index 3d3fb66..05b0195 100644
--- a/01.train_ODC_CNN.ipynb
+++ b/01.train_ODC_CNN.ipynb
@@ -2911,12 +2911,21 @@
     "# Import CNN model\n",
     "from cnn_model import CNNTrainer, reshape_for_cnn\n",
     "import torch\n",
+    "import numpy as np\n",
+    "\n",
+    "# Convert to numpy arrays if they're lists\n",
+    "X_train_np = np.array(X_train) if isinstance(X_train, list) else X_train.values\n",
+    "X_val_np = np.array(X_val) if isinstance(X_val, list) else X_val.values\n",
+    "X_test_np = np.array(X_test) if isinstance(X_test, list) else X_test.values\n",
+    "y_train_np = np.array(y_train) if isinstance(y_train, list) else y_train.values\n",
+    "y_val_np = np.array(y_val) if isinstance(y_val, list) else y_val.values\n",
+    "y_test_np = np.array(y_test) if isinstance(y_test, list) else y_test.values\n",
     "\n",
     "# Reshape data for CNN (n_samples, 39) -> (n_samples, 3, 13)\n",
     "print(\"🔄 Reshaping data for CNN...\")\n",
-    "X_train_cnn = reshape_for_cnn(X_train.values)\n",
-    "X_val_cnn = reshape_for_cnn(X_val.values)\n",
-    "X_test_cnn = reshape_for_cnn(X_test.values)\n",
+    "X_train_cnn = reshape_for_cnn(X_train_np)\n",
+    "X_val_cnn = reshape_for_cnn(X_val_np)\n",
+    "X_test_cnn = reshape_for_cnn(X_test_np)\n",
     "\n",
     "print(f\"   Train shape: {X_train_cnn.shape}\")\n",
     "print(f\"   Val shape: {X_val_cnn.shape}\")\n",
@@ -2948,8 +2957,8 @@
     "\n",
     "# Train model\n",
     "trainer.fit(\n",
-    "    X_train_cnn, y_train.values,\n",
-    "    X_val_cnn, y_val.values,\n",
+    "    X_train_cnn, y_train_np,\n",
+    "    X_val_cnn, y_val_np,\n",
     "    epochs=50,\n",
     "    batch_size=32,\n",
     "    verbose=True\n",
@@ -2979,7 +2988,7 @@
     "%%time\n",
     "# Evaluate on test data\n",
     "print(\"📊 Evaluating CNN model on test set...\\n\")\n",
-    "results = trainer.evaluate(X_test_cnn, y_test.values)\n",
+    "results = trainer.evaluate(X_test_cnn, y_test_np)\n",
     "\n",
     "# Plot confusion matrix\n",
     "from sklearn.metrics import ConfusionMatrixDisplay\n",