Merge branch 'main' of https://gitty.informatik.hs-mannheim.de/3016498/ANLP_WS24_CA2

BERT.py

@@ -54,12 +54,12 @@ if __name__ == '__main__':
     # Hyperparameter und Konfigurationen
     params = {
         # Training
-        "epochs": [1],
+        "epochs": [20],
         "patience": [7],
-        "learning_rate": [1e-5, 1e-6],
+        "learning_rate": [1e-5],
         "weight_decay": [5e-4],
         # Model
-        "dropout": [0.6]
+        "dropout": [0.3]
     }
 
     # Generate permutations of hyperparameters
@@ -67,9 +67,9 @@ if __name__ == '__main__':
     grid_params = [dict(zip(keys, v)) for v in itertools.product(*values)]
     best_params = {}
     best_params_rmse = -1
-    # Example usage of grid_params
-    # for param_set in grid_params:
-    #     print(param_set)
+    #Example usage of grid_params
+    for param_set in grid_params:
+        print(param_set)
     print('Number of grid_params:', len(grid_params))
 
 
@@ -81,18 +81,18 @@ if __name__ == '__main__':
     TEST_SIZE = 0.1
     VAL_SIZE = 0.1
 
-    MAX_LEN = 280
+    MAX_LEN = 128 #280
     BATCH_SIZE = 32
 
     N_MODELS = 1
-    USE_GIRD_SEARCH = True
+    USE_GIRD_SEARCH = False
 
     models = []
     timestamp = datetime.now().strftime("%Y-%m-%d_%H-%M-%S")
 
     # Daten laden und vorbereiten
-    embedding_matrix, word_index, vocab_size, d_model = dataset_helper.get_embedding_matrix(
-        gloVe_path=GLOVE_PATH, emb_len=EMBEDDING_DIM)
+    # embedding_matrix, word_index, vocab_size, d_model = dataset_helper.get_embedding_matrix(
+    #     gloVe_path=GLOVE_PATH, emb_len=EMBEDDING_DIM)
 
     X, y = dataset_helper.load_preprocess_data(path_data=DATA_PATH, verbose=True)
 
@@ -183,7 +183,7 @@ if __name__ == '__main__':
             test_r2 = r2_score(test_labels, test_preds)
             print(f"Test RMSE: {test_rmse:.4f}, Test MAE: {test_mae:.4f}, Test R²: {test_r2:.4f}")
 
-            if test_rmse > best_params_rmse:
+            if test_rmse < best_params_rmse:
                 best_params_rmse = test_rmse
                 best_params = params