diff --git a/fastNLP/core/callback.py b/fastNLP/core/callback.py
index b1a480cc..d941c235 100644
--- a/fastNLP/core/callback.py
+++ b/fastNLP/core/callback.py
@@ -248,7 +248,10 @@ class GradientClipCallback(Callback):
         self.clip_value = clip_value
 
     def on_backward_end(self, model):
-        self.clip_fun(model.parameters(), self.clip_value)
+        if self.parameters is None:
+            self.clip_fun(model.parameters(), self.clip_value)
+        else:
+            self.clip_fun(self.parameters, self.clip_value)
 
 
 class CallbackException(BaseException):
@@ -306,7 +309,6 @@ class LRScheduler(Callback):
 
     def on_epoch_begin(self, cur_epoch, total_epoch):
         self.scheduler.step()
-        print("scheduler step ", "lr=", self.trainer.optimizer.param_groups[0]["lr"])
 
 
 class ControlC(Callback):
diff --git a/fastNLP/modules/decoder/MLP.py b/fastNLP/modules/decoder/MLP.py
index c9198859..b76fdab7 100644
--- a/fastNLP/modules/decoder/MLP.py
+++ b/fastNLP/modules/decoder/MLP.py
@@ -7,7 +7,7 @@ from fastNLP.modules.utils import initial_parameter
 class MLP(nn.Module):
     """Multilayer Perceptrons as a decoder
 
-    :param list size_layer: list of int, define the size of MLP layers.
+    :param list size_layer: list of int, define the size of MLP layers. layer的层数为(len(size_layer)-1)//2 + 1
     :param str activation: str or function, the activation function for hidden layers.
     :param str initial_method: the name of initialization method.
     :param float dropout: the probability of dropout.