Fixing data paths

src/crazyneuraluser/UBAR_code/config.py

@@ -10,29 +10,25 @@ class _Config:
     def _multiwoz_ubar_init(self):
         self.gpt_path = "distilgpt2"
 
-        self.vocab_path_train = "cambridge-masters-project/data/preprocessed/UBAR/multi-woz-processed/vocab"
+        self.vocab_path_train = "data/preprocessed/UBAR/multi-woz-processed/vocab"
         self.vocab_path_eval = None
-        self.data_path = "cambridge-masters-project/data/preprocessed/UBAR/multi-woz-processed/"
+        self.data_path = "data/preprocessed/UBAR/multi-woz-processed/"
         self.data_file = "data_for_ubar.json"
-        self.dev_list = "cambridge-masters-project/data/raw/UBAR/multi-woz/valListFile.json"
-        self.test_list = "cambridge-masters-project/data/raw/UBAR/multi-woz/testListFile.json"
+        self.dev_list = "data/raw/UBAR/multi-woz/valListFile.json"
+        self.test_list = "data/raw/UBAR/multi-woz/testListFile.json"
         self.dbs = {
-            "attraction": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/attraction_db_processed.json",
-            "hospital": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/hospital_db_processed.json",
-            "hotel": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/hotel_db_processed.json",
-            "police": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/police_db_processed.json",
-            "restaurant": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/restaurant_db_processed.json",
-            "taxi": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/taxi_db_processed.json",
-            "train": "cambridge-masters-project/data/preprocessed/UBAR/db_processed/train_db_processed.json",
+            "attraction": "data/preprocessed/UBAR/db_processed/attraction_db_processed.json",
+            "hospital": "data/preprocessed/UBAR/db_processed/hospital_db_processed.json",
+            "hotel": "data/preprocessed/UBAR/db_processed/hotel_db_processed.json",
+            "police": "data/preprocessed/UBAR/db_processed/police_db_processed.json",
+            "restaurant": "data/preprocessed/UBAR/db_processed/restaurant_db_processed.json",
+            "taxi": "data/preprocessed/UBAR/db_processed/taxi_db_processed.json",
+            "train": "data/preprocessed/UBAR/db_processed/train_db_processed.json",
         }
-        self.glove_path = "cambridge-masters-project/data/glove/glove.6B.50d.txt"
-        self.domain_file_path = "cambridge-masters-project/data/preprocessed/UBAR/multi-woz-processed/domain_files.json"
-        self.slot_value_set_path = (
-            "cambridge-masters-project/data/preprocessed/UBAR/db_processed/value_set_processed.json"
-        )
-        self.multi_acts_path = (
-            "cambridge-masters-project/data/preprocessed/UBAR/multi-woz-processed/multi_act_mapping_train.json"
-        )
+        self.glove_path = "data/glove/glove.6B.50d.txt"
+        self.domain_file_path = "data/preprocessed/UBAR/multi-woz-processed/domain_files.json"
+        self.slot_value_set_path = "data/preprocessed/UBAR/db_processed/value_set_processed.json"
+        self.multi_acts_path = "data/preprocessed/UBAR/multi-woz-processed/multi_act_mapping_train.json"
         self.exp_path = "to be generated"
         self.log_time = time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime())
 

src/crazyneuraluser/UBAR_code/reader.py

@@ -301,7 +301,7 @@ class MultiWozReader(_ReaderBase):
                     self.exp_files[fn.replace(".json", "")] = 1
         #
 
-        self._load_data()
+        # self._load_data()
 
         if cfg.limit_bspn_vocab:
             self.bspn_masks = self._construct_bspn_constraint()
@@ -497,7 +497,7 @@ class MultiWozReader(_ReaderBase):
 
             if os.path.exists(encoded_file):
                 logging.info("Reading encoded data from {}".format(encoded_file))
-                self.data = json.loads(open(cfg.data_path + cfg.data_file, "r", encoding="utf-8").read().lower())
+                # self.data = json.loads(open(cfg.data_path + cfg.data_file, "r", encoding="utf-8").read().lower())
                 encoded_data = json.loads(open(encoded_file, "r", encoding="utf-8").read())
                 self.train = encoded_data["train"]
                 self.dev = encoded_data["dev"]
@@ -505,7 +505,7 @@ class MultiWozReader(_ReaderBase):
             else:
                 logging.info("Encoding data now and save the encoded data in {}".format(encoded_file))
                 # not exists, encode data and save
-                self.data = json.loads(open(cfg.data_path + cfg.data_file, "r", encoding="utf-8").read().lower())
+                # self.data = json.loads(open(cfg.data_path + cfg.data_file, "r", encoding="utf-8").read().lower())
                 self.train, self.dev, self.test = [], [], []
                 for fn, dial in self.data.items():
                     if ".json" in fn: