Black 20 release

src/transformers/data/datasets/glue.py

@@ -86,7 +86,10 @@ class GlueDataset(Dataset):
         cached_features_file = os.path.join(
             cache_dir if cache_dir is not None else args.data_dir,
             "cached_{}_{}_{}_{}".format(
-                mode.value, tokenizer.__class__.__name__, str(args.max_seq_length), args.task_name,
+                mode.value,
+                tokenizer.__class__.__name__,
+                str(args.max_seq_length),
+                args.task_name,
             ),
         )
         label_list = self.processor.get_labels()

src/transformers/data/datasets/language_modeling.py

@@ -21,7 +21,11 @@ class TextDataset(Dataset):
     """
 
     def __init__(
-        self, tokenizer: PreTrainedTokenizer, file_path: str, block_size: int, overwrite_cache=False,
+        self,
+        tokenizer: PreTrainedTokenizer,
+        file_path: str,
+        block_size: int,
+        overwrite_cache=False,
     ):
         assert os.path.isfile(file_path), f"Input file path {file_path} not found"
 
@@ -29,7 +33,12 @@ class TextDataset(Dataset):
 
         directory, filename = os.path.split(file_path)
         cached_features_file = os.path.join(
-            directory, "cached_lm_{}_{}_{}".format(tokenizer.__class__.__name__, str(block_size), filename,),
+            directory,
+            "cached_lm_{}_{}_{}".format(
+                tokenizer.__class__.__name__,
+                str(block_size),
+                filename,
+            ),
         )
 
         # Make sure only the first process in distributed training processes the dataset,

src/transformers/data/datasets/squad.py

@@ -119,7 +119,10 @@ class SquadDataset(Dataset):
         cached_features_file = os.path.join(
             cache_dir if cache_dir is not None else args.data_dir,
             "cached_{}_{}_{}_{}".format(
-                mode.value, tokenizer.__class__.__name__, str(args.max_seq_length), version_tag,
+                mode.value,
+                tokenizer.__class__.__name__,
+                str(args.max_seq_length),
+                version_tag,
             ),
         )
 

src/transformers/data/metrics/squad_metrics.py

@@ -589,10 +589,10 @@ def compute_predictions_log_probs(
     tokenizer,
     verbose_logging,
 ):
-    """ XLNet write prediction logic (more complex than Bert's).
-        Write final predictions to the json file and log-odds of null if needed.
+    """XLNet write prediction logic (more complex than Bert's).
+    Write final predictions to the json file and log-odds of null if needed.
 
-        Requires utils_squad_evaluate.py
+    Requires utils_squad_evaluate.py
     """
     _PrelimPrediction = collections.namedtuple(  # pylint: disable=invalid-name
         "PrelimPrediction", ["feature_index", "start_index", "end_index", "start_log_prob", "end_log_prob"]

src/transformers/data/processors/glue.py

@@ -69,7 +69,10 @@ def glue_convert_examples_to_features(
 if is_tf_available():
 
     def _tf_glue_convert_examples_to_features(
-        examples: tf.data.Dataset, tokenizer: PreTrainedTokenizer, task=str, max_length: Optional[int] = None,
+        examples: tf.data.Dataset,
+        tokenizer: PreTrainedTokenizer,
+        task=str,
+        max_length: Optional[int] = None,
     ) -> tf.data.Dataset:
         """
         Returns:

src/transformers/data/processors/utils.py

@@ -269,7 +269,9 @@ class SingleSentenceClassificationProcessor(DataProcessor):
                 logger.info("Tokenizing example %d", ex_index)
 
             input_ids = tokenizer.encode(
-                example.text_a, add_special_tokens=True, max_length=min(max_length, tokenizer.max_len),
+                example.text_a,
+                add_special_tokens=True,
+                max_length=min(max_length, tokenizer.max_len),
             )
             all_input_ids.append(input_ids)