Update to latest black (#708)

There is a new version of black, which is causing our format script
to fail. These are the Changes with the updated black script.

Author: mattdangerw

examples/bert_pretraining/bert_create_pretraining_data.py

@@ -124,7 +124,7 @@ def write_instance_to_example_files(
     writer = tf.io.TFRecordWriter(output_filename)
     total_written = 0
     lookup = dict(zip(vocab, range(len(vocab))))
-    for (inst_index, instance) in enumerate(instances):
+    for inst_index, instance in enumerate(instances):
         token_ids = [lookup[x] for x in instance.tokens]
         padding_mask = [1] * len(token_ids)
         segment_ids = list(instance.segment_ids)
@@ -379,7 +379,7 @@ def create_masked_lm_predictions(
     # (Issue #166)
 
     cand_indexes = []
-    for (i, token) in enumerate(tokens):
+    for i, token in enumerate(tokens):
         if token == "[CLS]" or token == "[SEP]":
             continue
         cand_indexes.append([i])

keras_nlp/layers/masked_lm_mask_generator.py

@@ -147,7 +147,11 @@ def call(self, inputs):
             # convert dense to ragged.
             inputs = tf.RaggedTensor.from_tensor(inputs)
 
-        (token_ids, mask_positions, mask_ids,) = tf_text.mask_language_model(
+        (
+            token_ids,
+            mask_positions,
+            mask_ids,
+        ) = tf_text.mask_language_model(
             inputs,
             item_selector=self._random_selector,
             mask_values_chooser=self._mask_values_chooser,

keras_nlp/metrics/bleu.py

@@ -223,7 +223,7 @@ def _corpus_bleu(
             smooth: boolean. Whether or not to apply Lin et al. 2004
                 smoothing.
         """
-        for (references, translation) in zip(
+        for references, translation in zip(
             reference_corpus, translation_corpus
         ):
             reference_length += min(len(r) for r in references)

keras_nlp/models/albert/albert_backbone.py

@@ -117,7 +117,6 @@ def __init__(
         num_segments=2,
         **kwargs,
     ):
-
         if num_layers % num_groups != 0:
             raise ValueError(
                 "`num_layers` must be divisible by `num_groups`. Received: "

keras_nlp/models/bart/bart_backbone.py

@@ -96,7 +96,6 @@ def __init__(
         max_sequence_length=1024,
         **kwargs,
     ):
-
         # Encoder inputs
         encoder_token_id_input = keras.Input(
             shape=(None,), dtype="int32", name="encoder_token_ids"

keras_nlp/models/bert/bert_backbone.py

@@ -104,7 +104,6 @@ def __init__(
         num_segments=2,
         **kwargs,
     ):
-
         # Index of classification token in the vocabulary
         cls_token_index = 0
         # Inputs

keras_nlp/models/deberta_v3/deberta_v3_backbone.py

@@ -108,7 +108,6 @@ def __init__(
         bucket_size=256,
         **kwargs,
     ):
-
         # Inputs
         token_id_input = keras.Input(
             shape=(None,), dtype="int32", name="token_ids"

keras_nlp/models/distil_bert/distil_bert_backbone.py

@@ -105,7 +105,6 @@ def __init__(
         max_sequence_length=512,
         **kwargs,
     ):
-
         # Inputs
         token_id_input = keras.Input(
             shape=(None,), dtype="int32", name="token_ids"

keras_nlp/models/f_net/f_net_backbone.py

@@ -100,7 +100,6 @@ def __init__(
         num_segments=4,
         **kwargs,
     ):
-
         # Index of classification token in the vocabulary
         cls_token_index = 0
         # Inputs

keras_nlp/models/gpt2/gpt2_backbone.py

@@ -102,7 +102,6 @@ def __init__(
         max_sequence_length=1024,
         **kwargs,
     ):
-
         # Inputs
         token_ids = keras.Input(shape=(None,), dtype="int32", name="token_ids")
         padding_mask = keras.Input(

keras_nlp/models/gpt2/gpt2_preprocessor.py

@@ -121,7 +121,6 @@ def __init__(
         sequence_length,
         **kwargs,
     ):
-
         super().__init__(**kwargs)
 
         self.tokenizer = tokenizer

keras_nlp/models/roberta/roberta_backbone.py

@@ -98,7 +98,6 @@ def __init__(
         max_sequence_length=512,
         **kwargs,
     ):
-
         # Inputs
         token_id_input = keras.Input(
             shape=(None,), dtype=tf.int32, name="token_ids"

keras_nlp/samplers/beam_sampler.py

@@ -111,7 +111,6 @@ def sample(
         )
 
         def one_step(beams, beams_prob, length, mask):
-
             flattened_beams = tf.reshape(
                 beams, shape=[batch_size * num_beams, -1]
             )

keras_nlp/tokenizers/byte_tokenizer.py

@@ -207,7 +207,6 @@ def vocabulary_size(self) -> int:
         return 256
 
     def tokenize(self, inputs):
-
         if not isinstance(inputs, (tf.Tensor, tf.RaggedTensor)):
             inputs = tf.convert_to_tensor(inputs)
 

keras_nlp/tokenizers/byte_tokenizer_test.py

@@ -223,7 +223,6 @@ def test_load_model_with_config(self):
         )
 
     def test_config(self):
-
         tokenizer = ByteTokenizer(
             name="byte_tokenizer_config_test",
             lowercase=False,