yapf format and solve travis-ci problem

Author: JingqingZ

tensorlayer/layers/recurrent.py

@@ -105,7 +105,10 @@ class RNN(Layer):
     Similar to the DynamicRNN in TL 1.x.
 
     If the `sequence_length` is provided in RNN's forwarding and both `return_last_output` and `return_last_state`
-    are set as `True`, the forward function will automatically ignore the paddings.
+    are set as `True`, the forward function will automatically ignore the paddings. Note that if `return_last_output`
+    is set as `False`, the synced sequence outputs will still include outputs which correspond with paddings,
+    but users are free to select which slice of outputs to be used in following procedure.
+
     The `sequence_length` should be a list of integers which indicates the length of each sequence.
     It is recommended to
     `tl.layers.retrieve_seq_length_op3 <https://tensorlayer.readthedocs.io/en/latest/modules/layers.html#compute-sequence-length-3>`__
@@ -1074,6 +1077,7 @@ def __init__(
 
 '''
 
+
 # @tf.function
 def retrieve_seq_length_op(data):
     """An op to compute the length of a sequence from input shape of [batch_size, n_step(max), n_features],

tests/layers/test_layers_recurrent.py

@@ -18,7 +18,7 @@ class Layer_RNN_Test(CustomTestCase):
     @classmethod
     def setUpClass(cls):
 
-        cls.batch_size = 10
+        cls.batch_size = 2
 
         cls.vocab_size = 20
         cls.embedding_size = 4
@@ -878,9 +878,7 @@ class CustomisedModel(tl.models.Model):
             def __init__(self):
                 super(CustomisedModel, self).__init__()
                 self.rnnlayer = tl.layers.LSTMRNN(
-                    units=8, dropout=0.1, in_channels=4,
-                    return_last_output=True,
-                    return_last_state=False
+                    units=8, dropout=0.1, in_channels=4, return_last_output=True, return_last_state=False
                 )
                 self.dense = tl.layers.Dense(in_channels=8, n_units=1)
 
@@ -924,7 +922,6 @@ def forward(self, x):
         os.remove(filename)
 
 
-
 if __name__ == '__main__':
 
     unittest.main()