fix(archai): Fixes functions that call get_full_path with create_folder argument.

archai/datasets/nlp/nvidia_dataset_provider_utils.py

@@ -59,7 +59,7 @@ class Corpus:
 
         # Corpus cache is created using dataset/vocab_type/vocab_size path
         self.corpus_cache_dir = get_full_path(
-            os.path.join(cache_dir, str(dataset_name), str(vocab_type), str(vocab_size)), create=True
+            os.path.join(cache_dir, str(dataset_name), str(vocab_type), str(vocab_size)), create_folder=True
         )
 
         # Encoded dataset (.npy files) cache paths

archai/datasets/nlp/tokenizer_utils/bbpe_tokenizer.py

@@ -72,7 +72,7 @@ class BbpeTokenizer(TokenizerBase):
         )
 
         self._tokenizer = None
-        self._tokenizer_filepath = os.path.join(get_full_path(save_path, create=True), "bbpe_tokenizer.json")
+        self._tokenizer_filepath = os.path.join(get_full_path(save_path, create_folder=True), "bbpe_tokenizer.json")
 
         self.vocab_size = vocab_size
         self.sorted_vocab = sorted_vocab

archai/datasets/nlp/tokenizer_utils/word_tokenizer.py

@@ -204,7 +204,7 @@ class WordTokenizer(TokenizerBase):
         self.sym2idx = OrderedDict()
 
     def _vocab_filepath(self) -> str:
-        vocab_dir = get_full_path(os.path.join(self.save_path), create=True)
+        vocab_dir = get_full_path(os.path.join(self.save_path), create_folder=True)
 
         return os.path.join(vocab_dir, "vocab.txt")
 

archai/trainers/nlp/nvidia_trainer.py

@@ -192,14 +192,14 @@ class NvidiaTrainer(TrainerBase):
         else:
             raise RuntimeError(f"Split: {split} is not supported yet.")
 
-        if self.dataset_name in ["wt2", "wt103"] or self.dataset_name.startswith("olx_"):
+        if self.args.dataset_name in ["wt2", "wt103"] or self.args.dataset_name.startswith("olx_"):
             return LMOrderedIterator(
                 input_ids,
                 self.args.global_batch_size,
                 self.args.seq_len,
                 device=self.args.device,
             )
-        elif self.dataset_name == "lm1b":
+        elif self.args.dataset_name == "lm1b":
             return LMMultiFileIterator(
                 input_ids,
                 self.vocab,
@@ -208,7 +208,7 @@ class NvidiaTrainer(TrainerBase):
                 device=self.args.device,
             )
         else:
-            raise RuntimeError(f"Dataset: {self.dataset_name} is not supported yet.")
+            raise RuntimeError(f"Dataset: {self.args.dataset_name} is not supported yet.")
 
     def _create_optimizer(self) -> None:
         optimizer_name = self.args.optim.lower()

scripts/trainers/nlp/train_with_nvidia_trainer.py

@@ -20,9 +20,9 @@ def parse_args() -> argparse.Namespace:
 
     parser.add_argument("-es", "--eval_steps", type=int, default=100, help="Number of steps between evaluations.")
 
-    parser.add_argument("-d", "--dataset", type=str, default="wt103", help="Name of the dataset.")
+    parser.add_argument("-dn", "--dataset_name", type=str, default="wt103", help="Name of the dataset.")
 
-    parser.add_argument("-v", "--vocab", type=str, default="gpt2", help="Name of the vocabulary/tokenizer.")
+    parser.add_argument("-vt", "--vocab_type", type=str, default="gpt2", help="Name of the vocabulary/tokenizer.")
 
     parser.add_argument("-vs", "--vocab_size", type=int, default=10000, help="Size of the vocabulary.")
 
@@ -54,8 +54,8 @@ if __name__ == "__main__":
         no_cuda=args.no_cuda,
         logging_steps=args.logging_steps,
         eval_steps=args.eval_steps,
-        dataset=args.dataset,
+        dataset_name=args.dataset_name,
-        vocab=args.vocab,
+        vocab_type=args.vocab_type,
         vocab_size=args.vocab_size,
         global_batch_size=args.global_batch_size,
         seq_len=args.seq_len,