diff --git a/.gitignore b/.gitignore
index ebe097c2a7..b34cf27262 100755
--- a/.gitignore
+++ b/.gitignore
@@ -9,4 +9,5 @@ Session.vim
 .idea
 .vscode
 __pycache__
+.pytest*
 venv
\ No newline at end of file
diff --git a/setup.cfg b/setup.cfg
index 84607c1876..5921cbcd9c 100755
--- a/setup.cfg
+++ b/setup.cfg
@@ -1,8 +1,8 @@
 [flake8]
 ignore = E402,E701,E702,E704,E251
-max-line-length = 150
+max-line-length = 127
 
 [pep8]
 ignore = E402,E701,E702,E704,E251
-max-line-length = 150
+max-line-length = 127
 indent-size = 4
diff --git a/tensorflow_asr/models/streaming_transducer.py b/tensorflow_asr/models/streaming_transducer.py
index 60146f4e9d..38e13a3f77 100644
--- a/tensorflow_asr/models/streaming_transducer.py
+++ b/tensorflow_asr/models/streaming_transducer.py
@@ -257,7 +257,7 @@ def recognize(self, signals):
         """
         def execute(signal: tf.Tensor):
             features = self.speech_featurizer.tf_extract(signal)
-            encoded, _ = self.encoder_inference(features, self.encoder.get_initial_states())
+            encoded, _ = self.encoder_inference(features, self.encoder.get_initial_state())
             hypothesis = self.perform_greedy(
                 encoded,
                 predicted=tf.constant(self.text_featurizer.blank, dtype=tf.int32),
@@ -310,10 +310,13 @@ def recognize_beam(self, signals, lm=False):
         """
         def execute(signal: tf.Tensor):
             features = self.speech_featurizer.tf_extract(signal)
-            encoded, _ = self.encoder_inference(features, self.encoder.get_initial_states())
+            encoded, _ = self.encoder_inference(features, self.encoder.get_initial_state())
             hypothesis = self.perform_beam_search(encoded, lm)
-            prediction = tf.map_fn(lambda x: tf.strings.to_number(x, tf.int32),
-                                   tf.strings.split(hypothesis.prediction), fn_output_signature=tf.TensorSpec([], dtype=tf.int32))
+            prediction = tf.map_fn(
+                lambda x: tf.strings.to_number(x, tf.int32),
+                tf.strings.split(hypothesis.prediction),
+                fn_output_signature=tf.TensorSpec([], dtype=tf.int32)
+            )
             transcripts = self.text_featurizer.iextract(tf.expand_dims(prediction, axis=0))
             return tf.squeeze(transcripts)  # reshape from [1] to []
 
diff --git a/tensorflow_asr/models/transducer.py b/tensorflow_asr/models/transducer.py
index bc54691b9b..175847ac09 100755
--- a/tensorflow_asr/models/transducer.py
+++ b/tensorflow_asr/models/transducer.py
@@ -451,8 +451,11 @@ def execute(signal: tf.Tensor):
             features = self.speech_featurizer.tf_extract(signal)
             encoded = self.encoder_inference(features)
             hypothesis = self.perform_beam_search(encoded, lm)
-            prediction = tf.map_fn(lambda x: tf.strings.to_number(x, tf.int32),
-                                   tf.strings.split(hypothesis.prediction), fn_output_signature=tf.TensorSpec([], dtype=tf.int32))
+            prediction = tf.map_fn(
+                lambda x: tf.strings.to_number(x, tf.int32),
+                tf.strings.split(hypothesis.prediction),
+                fn_output_signature=tf.TensorSpec([], dtype=tf.int32)
+            )
             transcripts = self.text_featurizer.iextract(tf.expand_dims(prediction, axis=0))
             return tf.squeeze(transcripts)  # reshape from [1] to []
 
diff --git a/tests/plot_learning_rate.py b/tests/plot_learning_rate.py
index e44196a449..96bcbad620 100755
--- a/tests/plot_learning_rate.py
+++ b/tests/plot_learning_rate.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import math
 import tensorflow as tf
 import matplotlib.pyplot as plt
 from tensorflow_asr.optimizers.schedules import SANSchedule, TransformerSchedule