tokenizer.save_pretrained: only save file if non-empty

src/transformers/configuration_auto.py

@@ -12,7 +12,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-""" Auto Model class. """
+""" Auto Config class. """
 
 
 import logging

src/transformers/tokenization_utils.py

@@ -513,11 +513,9 @@ class PreTrainedTokenizer(object):
         with open(special_tokens_map_file, "w", encoding="utf-8") as f:
             f.write(json.dumps(self.special_tokens_map, ensure_ascii=False))
 
+        if len(self.added_tokens_encoder) > 0:
             with open(added_tokens_file, "w", encoding="utf-8") as f:
-            if self.added_tokens_encoder:
                 out_str = json.dumps(self.added_tokens_encoder, ensure_ascii=False)
-            else:
-                out_str = "{}"
                 f.write(out_str)
 
         vocab_files = self.save_vocabulary(save_directory)

tests/test_tokenization_auto.py

@@ -33,13 +33,13 @@ class AutoTokenizerTest(unittest.TestCase):
     # @slow
     def test_tokenizer_from_pretrained(self):
         logging.basicConfig(level=logging.INFO)
-        for model_name in [x for x in BERT_PRETRAINED_CONFIG_ARCHIVE_MAP.keys() if "japanese" not in x]:
+        for model_name in (x for x in BERT_PRETRAINED_CONFIG_ARCHIVE_MAP.keys() if "japanese" not in x):
             tokenizer = AutoTokenizer.from_pretrained(model_name)
             self.assertIsNotNone(tokenizer)
             self.assertIsInstance(tokenizer, BertTokenizer)
             self.assertGreater(len(tokenizer), 0)
 
-        for model_name in list(GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP.keys())[:1]:
+        for model_name in GPT2_PRETRAINED_CONFIG_ARCHIVE_MAP.keys():
             tokenizer = AutoTokenizer.from_pretrained(model_name)
             self.assertIsNotNone(tokenizer)
             self.assertIsInstance(tokenizer, GPT2Tokenizer)