diff --git a/draft_model_5.py b/draft_model_5.py
index 652b749b705cf448282755648e38c82cd7a32071..4b8edc9275a9c1c8d2748f3a897c4c34a48f6dcf 100644
--- a/draft_model_5.py
+++ b/draft_model_5.py
@@ -157,9 +157,6 @@ x = LSTM(units=200)(x)
 x = BatchNormalization()(x)
 output_lvl_1 = LeakyReLU(alpha=0.3)(x)
 
-# x = LeakyReLU(alpha=0.3)(x)
-# output_lvl_1 = Dense(20, activation='sigmoid')(x)
-
 model_lvl_1 = Model(inputs=input_lvl_1, outputs=output_lvl_1)
 
 
@@ -183,16 +180,17 @@ main_output = Dense(9, activation='sigmoid')(x)
 main_model = Model(inputs=main_input, outputs=main_output)
 main_model.summary()
 
-try:
-    model = multi_gpu_model(main_model, gpus=4, cpu_relocation=False)
-    print("Training on 4 GPUs")
-except:
-    print("Training on 1 GPU (or CPU)")
+# try:
+#     model = multi_gpu_model(main_model, gpus=4, cpu_relocation=False)
+#     print("Training on 4 GPUs")
+# except:
+#     print("Training on 1 GPU (or CPU)")
 
 main_model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['categorical_accuracy'])
 main_model.summary()
 
 
+##########################################################
 main_model.fit(arr_seg_train, arr_labels_train, epochs=20, batch_size=24,
           verbose=1, validation_data=(arr_seg_test, arr_labels_test), shuffle=True)