compute sent vecs mit best model

scripts/ActorOnly+CEL/main_ActorOnly.py

@@ -54,6 +54,13 @@ for epoch in range(epochs):
         # (over)write best model in file, and for Critic: save model_actor_only   
         torch.save(m.state_dict(), 'model_actor_only_wts.pth')
 
+        # load best model and compute sent_vecs anew
+        best_m = ActorOnlySummarisationModel()
+        best_m.load_state_dict(best_model_wts)
+        train_data.compute_sent_vecs(best_m, 'workspace/students/kreuzer/new_train')  #
+        val_data.compute_sent_vecs(best_m, 'workspace/students/kreuzer/new_val')
+        test_data.compute_sent_vecs(best_m, 'workspace/students/kreuzer/new_test')
+
     time_elapsed = time.time() - since
     print('Training + validation running already {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))
     print('Best val rouge so far: {:4f}'.format(best_rouge))
@@ -72,14 +79,6 @@ print('Best val rouge: {:4f}'.format(best_rouge))
 
 
 
-# load best model weights
-m.load_state_dict(best_model_wts)
-
-
-# set sent_vecs 
-train_data.compute_sent_vecs(m, 'workspace/students/kreuzer/new_train')  #
-val_data.compute_sent_vecs(m, 'workspace/students/kreuzer/new_val')
-test_data.compute_sent_vecs(m, 'workspace/students/kreuzer/new_test')
 
 # testing
 since = time.time()

structures.py

@@ -92,7 +92,7 @@ class PreprocessedDataSet:
             self.path = path
         
             self.path.mkdir()
-                
+            
             self.length = len(dataset)
             
             for i, element in enumerate(dataset):
@@ -132,8 +132,9 @@ class PreprocessedDataSet:
         
     def compute_sent_vecs(self, model, path):
         
-        path = pathlib.Path(path)
-        path.mkdir()
+        # writes in file, allows overwriting in main_ActorOnly
+        pathlib.Path(path).mkdir(parents=True, exist_ok=True)
+
         for i in range(self.length):
             dp = self.__getitem__(i)
             dp.compute_sent_vecs(model)