diff --git a/keras_transformer/attention.py b/keras_transformer/attention.py
index ee6cb83..598c04d 100644
--- a/keras_transformer/attention.py
+++ b/keras_transformer/attention.py
@@ -36,7 +36,7 @@ def __init__(self, num_heads: int, use_masking: bool,
         if (compression_window_size is not None
                 and compression_window_size <= 0):
             assert ValueError(
-                f"Too small compression window ({compression_window_size})")
+                "Too small compression window ({})".format(compression_window_size))
         self.compression_window_size = compression_window_size
         super().__init__(**kwargs)
 
@@ -83,10 +83,7 @@ def build_output_params(self, d_model):
 
     def validate_model_dimensionality(self, d_model: int):
         if d_model % self.num_heads != 0:
-            raise ValueError(
-                f'The size of the last dimension of the input '
-                f'({d_model}) must be evenly divisible by the number'
-                f'of the attention heads {self.num_heads}')
+            raise ValueError('The size of the last dimension of the input ({}) must be evenly divisible by the number of the attention heads {}'.format([d_model,self.num_heads]))
 
     def attention(self, pre_q, pre_v, pre_k, out_seq_len: int, d_model: int,
                   training=None):
@@ -233,10 +230,7 @@ def build(self, input_shape):
                 '(for keys/values and queries)')
         values_dim, query_dim = input_shape[0][-1], input_shape[1][-1]
         if query_dim != values_dim:
-            raise ValueError(
-                f'Both keys/value and query inputs must be '
-                f'of the same dimensionality, instead of '
-                f'{values_dim} and {query_dim}.')
+            raise ValueError('Both keys/value and query inputs must be of the same dimensionality, instead of {} and {}.'.format([values_dim,query_dim]))
         d_model = query_dim
         self.validate_model_dimensionality(d_model)
         # These weights are concatenated matrices W_k and W_v which