diff --git a/paddlehub/module/module.py b/paddlehub/module/module.py
index b73c55332ac290fc65808bbd7ca5105a6bcbc0b4..39d7e63ff58954267829c2fb166f6c746dfb3537 100644
--- a/paddlehub/module/module.py
+++ b/paddlehub/module/module.py
@@ -257,8 +257,8 @@ class Module(fluid.dygraph.Layer):
     def _initialize(self):
         pass
 
-    def forward(self, *args):
-        return self.model_runner(*args)
+    def forward(self, *args, **kwargs):
+        return self.model_runner(*args, **kwargs)
 
 
 class ModuleHelper(object):
diff --git a/paddlehub/module/nlp_module.py b/paddlehub/module/nlp_module.py
index f99a672d204dca47285c7ff452c83bc9ca64a8a8..c7b673717ed7a24707d3aac93f2a9e1d587bb388 100644
--- a/paddlehub/module/nlp_module.py
+++ b/paddlehub/module/nlp_module.py
@@ -353,6 +353,13 @@ class TransformerModule(NLPBaseModule):
 
         return inputs, outputs, module_program
 
+    @property
+    def model_runner(self):
+        if not self._model_runner:
+            self._model_runner = fluid.dygraph.StaticModelRunner(
+                self.params_path)
+        return self._model_runner
+
     def get_embedding(self, texts, use_gpu=False, batch_size=1):
         """
         get pooled_output and sequence_output for input texts.