diff --git a/pytorch_transformers/modeling_bert.py b/pytorch_transformers/modeling_bert.py
index 7b34b3fd..badec992 100644
--- a/pytorch_transformers/modeling_bert.py
+++ b/pytorch_transformers/modeling_bert.py
@@ -216,7 +216,7 @@ class BertConfig(PretrainedConfig):
             self.layer_norm_eps = layer_norm_eps
         else:
             raise ValueError("First argument must be either a vocabulary size (int)"
-                             "or the path to a pretrained model config file (str)")
+                             " or the path to a pretrained model config file (str)")
 
 
 
diff --git a/pytorch_transformers/modeling_transfo_xl.py b/pytorch_transformers/modeling_transfo_xl.py
index 3cfdee38..c57e664c 100644
--- a/pytorch_transformers/modeling_transfo_xl.py
+++ b/pytorch_transformers/modeling_transfo_xl.py
@@ -285,7 +285,7 @@ class TransfoXLConfig(PretrainedConfig):
             self.init_std = init_std
         else:
             raise ValueError("First argument must be either a vocabulary size (int)"
-                             "or the path to a pretrained model config file (str)")
+                             " or the path to a pretrained model config file (str)")
 
     @property
     def max_position_embeddings(self):
diff --git a/pytorch_transformers/modeling_xlm.py b/pytorch_transformers/modeling_xlm.py
index 19800da2..5a659e02 100644
--- a/pytorch_transformers/modeling_xlm.py
+++ b/pytorch_transformers/modeling_xlm.py
@@ -178,7 +178,7 @@ class XLMConfig(PretrainedConfig):
             self.end_n_top = end_n_top
         else:
             raise ValueError("First argument must be either a vocabulary size (int)"
-                             "or the path to a pretrained model config file (str)")
+                             " or the path to a pretrained model config file (str)")
 
     @property
     def vocab_size(self):
diff --git a/pytorch_transformers/modeling_xlnet.py b/pytorch_transformers/modeling_xlnet.py
index d4482178..136f07c4 100644
--- a/pytorch_transformers/modeling_xlnet.py
+++ b/pytorch_transformers/modeling_xlnet.py
@@ -306,7 +306,7 @@ class XLNetConfig(PretrainedConfig):
             self.end_n_top = end_n_top
         else:
             raise ValueError("First argument must be either a vocabulary size (int)"
-                             "or the path to a pretrained model config file (str)")
+                             " or the path to a pretrained model config file (str)")
 
     @property
     def max_position_embeddings(self):