diff --git a/.gitignore b/.gitignore
index a98485a..fc96cd3 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,3 +1,4 @@
 .vscode
 __pycache__
-data/
\ No newline at end of file
+data/
+results/
\ No newline at end of file
diff --git a/src/evaluate.py b/src/evaluate.py
index 0877170..052a7ce 100644
--- a/src/evaluate.py
+++ b/src/evaluate.py
@@ -15,7 +15,7 @@ def evaluate(
     model: nn.Module,
     test_data: pd.DataFrame,
     batch_size: int,
-) -> None:
+) -> list[int]:
     test_dataset = Dataset(test_data)
 
     test_dataloader = torch.utils.data.DataLoader(
@@ -26,8 +26,11 @@ def evaluate(
         shuffle=True,
     )
 
-    model.to(DEVICE)
     total_acc_test = 0
+    results = []
+
+    model.to(DEVICE)
+    model.eval()
 
     with torch.no_grad():
         for test_input, test_label in test_dataloader:
@@ -38,10 +41,8 @@ def evaluate(
             output = model(input_id, mask)
 
             acc = (output.argmax(dim=1) == test_label).sum().item()
+            results.extend(output.argmax(dim=1).tolist())
             total_acc_test += acc
 
     print(f"Test Accuracy: {total_acc_test / len(test_data): .3f}")
-
-
-if __name__ == "__main__":
-    pass
+    return results
diff --git a/src/main.py b/src/main.py
index 39ceda9..b48d32c 100644
--- a/src/main.py
+++ b/src/main.py
@@ -1,7 +1,8 @@
 import random
 from sklearn.model_selection import train_test_split
+import argparse
 
-from models import BertClassifier
+from models import BertClassifier, utils
 from datasets import NewsDataset
 from train import train
 from evaluate import evaluate
@@ -11,38 +12,60 @@ SEED = 2137
 # Hyperparameters
 
 INITIAL_LR = 1e-6
-NUM_EPOCHS = 5
+NUM_EPOCHS = 3
 BATCH_SIZE = 2
 
+
+# argument parser
+
+parser = argparse.ArgumentParser(
+    prog="News classification",
+    description="Train or evaluate model",
+)
+parser.add_argument("--train", action="store_true", default=False)
+parser.add_argument("--test", action="store_true", default=False)
+parser.add_argument("--model_path", type=str, default="results/model.pt")
+parser.add_argument("--results_path", type=str, default="results/results.csv")
+
+
 if __name__ == "__main__":
+    args = parser.parse_args()
+
     # loading & spliting data
     news_dataset = NewsDataset(data_dir_path="data", data_lenght=2000)
 
     train_val_data, test_data = train_test_split(
         news_dataset.data,
-        test_size=0.8,
+        test_size=0.2,
         shuffle=True,
         random_state=random.seed(SEED),
     )
+
     train_data, val_data = train_test_split(
         train_val_data,
         test_size=0.2,
         shuffle=True,
         random_state=random.seed(SEED),
     )
+
     # trainig model
-    trained_model = train(
-        model=BertClassifier(),
-        train_data=train_data,
-        val_data=val_data,
-        learning_rate=INITIAL_LR,
-        epochs=NUM_EPOCHS,
-        batch_size=BATCH_SIZE,
-    )
+    if args.train:
+        trained_model = train(
+            model=BertClassifier(),
+            train_data=test_data,
+            val_data=val_data,
+            learning_rate=INITIAL_LR,
+            epochs=NUM_EPOCHS,
+            batch_size=BATCH_SIZE,
+        )
+        utils.save_model(model=trained_model, model_path=args.model_path)
 
     # evaluating model
-    evaluate(
-        model=trained_model,
-        test_data=test_data,
-        batch_size=BATCH_SIZE,
-    )
+    if args.test:
+        model = utils.load_model(model=BertClassifier(), model_path=args.model_path)  # loading model from model.pt file
+        results = evaluate(
+            model=model,
+            test_data=test_data,
+            batch_size=BATCH_SIZE,
+        )
+        utils.save_results(labels=test_data["label"], results=results, file_path=args.results_path)
diff --git a/src/models/__init__.py b/src/models/__init__.py
index fed70df..d255641 100644
--- a/src/models/__init__.py
+++ b/src/models/__init__.py
@@ -1,3 +1,4 @@
-__all__ = ["BertClassifier"]
+__all__ = ["BertClassifier", "utils"]
 
 from .bert_model import BertClassifier
+from .utils import utils
diff --git a/src/models/utils.py b/src/models/utils.py
new file mode 100644
index 0000000..ca5465f
--- /dev/null
+++ b/src/models/utils.py
@@ -0,0 +1,34 @@
+from pathlib import Path
+
+import torch
+import torch.nn as nn
+import pandas as pd
+
+
+class utils:
+    def __init__(self) -> None:
+        pass
+
+    @staticmethod
+    def save_model(model: nn.Module, model_path: str) -> None:
+        model_path = Path(model_path)
+        model_path.parent.mkdir(parents=True, exist_ok=True)
+
+        torch.save(model.state_dict(), model_path)
+        print(f"[INFO]\t Model saved at: {model_path}")
+
+    @staticmethod
+    def load_model(model: nn.Module, model_path: str) -> nn.Module:
+        model_path = Path(model_path)
+        model_path.parent.mkdir(parents=True, exist_ok=True)
+
+        model.load_state_dict(torch.load(model_path))
+        return model
+
+    @staticmethod
+    def save_results(labels: list[int], results: list[int], file_path: str) -> None:
+        file_path = Path(file_path)
+        file_path.parent.mkdir(parents=True, exist_ok=True)
+
+        df = pd.DataFrame({"labels": labels, "results": results})
+        df.to_csv(file_path, index=False)