diff --git a/AsyncDataLoader.py b/AsyncDataLoader.py
index 3bdb07f..ce484f3 100644
--- a/AsyncDataLoader.py
+++ b/AsyncDataLoader.py
@@ -51,7 +51,6 @@ class AsyncDataLoader(torch.utils.data.DataLoader):
 
                     self.__dataset_access.acquire()
 
-
         self.__dataset_access.release()
 
 
diff --git a/cnn_train.py b/cnn_train.py
index 8b02705..326bef2 100644
--- a/cnn_train.py
+++ b/cnn_train.py
@@ -4,7 +4,7 @@ import numpy.random
 import torch.utils.data
 import torch.cuda
 
-from architecture import MyCNN
+from architecture import model
 from dataset import ImagesDataset
 
 from AImageDataset import AImagesDataset
@@ -49,10 +49,7 @@ def train_model(accuracies,
 
     dataset = ImagesDataset("training_data")
 
-    # dataset = torch.utils.data.Subset(dataset, range(0, 32))
-
     train_data, eval_data = split_data(dataset)
-    # train_data, eval_data = torch.utils.data.random_split(dataset, [0.5, 0.5])
 
     augmented_train_data = AImagesDataset(train_data, augment_data)
     train_loader = AsyncDataLoader(augmented_train_data,
@@ -123,12 +120,8 @@ def train_model(accuracies,
         losses.append('eval_loss', eval_loss.item() / len(eval_data))
         print("Eval: ", eval_positives.item(), "/ ", len(eval_data), " = ", eval_positives.item() / len(eval_data))
 
-        # print epoch summary
-
-        # print(f"Epoch: {epoch} --- Train loss: {train_loss:7.4f} --- Eval loss: {eval_loss:7.4f}")
-
         if eval_positives.item() / len(eval_data) > 0.5:
-            torch.save(model.state_dict(), f'models/model-{start_time.strftime("%Y%m%d-%H%M%S")}-epoch-{epoch}.pt')
+            torch.save(model.state_dict(), f'models/model-{start_time.strftime("%Y%m%d-%H%M%S")}-epoch-{epoch}.pth')
         with open(f'models/model-{start_time.strftime("%Y%m%d-%H%M%S")}.csv', 'a') as file:
             file.write(f'{epoch};{len(augmented_train_data)};{len(eval_data)};{train_loss.item()};{eval_loss.item()};'
                        f'{train_positives};{eval_positives}\n')
@@ -140,15 +133,13 @@ def train_worker(p_epoch, p_train, p_eval, plotter_accuracies, plotter_loss, sta
 
     device = 'cuda'
 
-    model = MyCNN(input_channels=1,
-                  input_size=(100, 100)).to(device)
+    model.to(device)
 
     num_epochs = 1000000
 
     batch_size = 64
     optimizer = torch.optim.Adam(model.parameters(),
                                  lr=0.0001,
-                                 # weight_decay=0.1,
                                  fused=True)
     loss_function = torch.nn.CrossEntropyLoss()
 
diff --git a/verify.py b/verify.py
new file mode 100644
index 0000000..ebb2784
--- /dev/null
+++ b/verify.py
@@ -0,0 +1,26 @@
+import torch
+import torch.utils.data
+
+from architecture import model
+from dataset import ImagesDataset
+
+if __name__ == "__main__":
+    model_params = torch.load("submit_attempts/Nr1/model-20240713-164545-epoch-30.pth")
+
+    model.load_state_dict(model_params)
+    model.eval()
+
+    dataset = ImagesDataset("training_data")
+
+    correct = 0
+
+    print("evaluating...")
+
+    for (image_t, class_id, _, _) in torch.utils.data.DataLoader(dataset):
+        out = model(image_t)
+
+        if out.argmax() == class_id:
+            correct += 1
+
+    print(f"Identified {correct} images out of {len(dataset)} correctly")
+    print(f"Accuracy: {100 * correct / len(dataset)}%")