diff --git a/paddle/utils/Util.cpp b/paddle/utils/Util.cpp
index 2cdff9d1aca927122fcdb0c2a7ab22a0e38b41c1..b16d4314654ffeab74137ec1ee69203dab56d851 100644
--- a/paddle/utils/Util.cpp
+++ b/paddle/utils/Util.cpp
@@ -378,7 +378,7 @@ hl_activation_mode_t hlActiveType(const std::string& type) {
     return HL_ACTIVATION_RELU;
   } else if (type == "tanh") {
     return HL_ACTIVATION_TANH;
-  } else if (type == "linear") {
+  } else if (type == "linear" || type == "") {
     return HL_ACTIVATION_LINEAR;
   } else {
     LOG(FATAL) << "Do not support activation type " << type;
diff --git a/python/paddle/trainer/config_parser.py b/python/paddle/trainer/config_parser.py
index d11ee1feca62ef53079291d93de7333991b65df7..92bf2920b8996c75796e4f5eaeee6f8508fdf79e 100644
--- a/python/paddle/trainer/config_parser.py
+++ b/python/paddle/trainer/config_parser.py
@@ -218,7 +218,7 @@ def Inputs(*args):
 
 @config_func
 def HasInputsSet():
-    return len(g_config.model_config.input_layer_names) != 0
+    return len(g_current_submodel.input_layer_names) != 0
 
 
 # Define the name of the output layers of the NeuralNetwork.
@@ -1170,14 +1170,14 @@ def parse_block_expand(block_expand, input_layer_name, block_expand_conf):
         block_expand_conf.output_x = 0
     else:
         block_expand_conf.output_x = cnn_output_size(
-            block_expand.img_size_x, block_expand.block_x, 
+            block_expand.img_size_x, block_expand.block_x,
             block_expand.padding_x, block_expand.stride_x, False)
 
     if block_expand_conf.img_size_y == 0:
         block_expand_conf.output_y = 0
     else:
         block_expand_conf.output_y = cnn_output_size(
-            block_expand.img_size_y, block_expand.block_y, 
+            block_expand.img_size_y, block_expand.block_y,
             block_expand.padding_y, block_expand.stride_y, False)
 
 def parse_maxout(maxout, input_layer_name, maxout_conf):
@@ -1185,7 +1185,7 @@ def parse_maxout(maxout, input_layer_name, maxout_conf):
     maxout_conf.groups = maxout.groups
     maxout_conf.img_size_x = maxout.img_size_x
     maxout_conf.img_size_y = maxout.img_size_y
-    
+
 # Define an evaluator
 @config_func
 def Evaluator(
@@ -1881,7 +1881,7 @@ class MaxOutLayer(LayerBase):
                      self.config.inputs[0].maxout_conf)
         maxout_conf = self.config.inputs[0].maxout_conf
         self.set_layer_size(g_layer_map[input_layer.name].size / maxout_conf.groups)
-            
+
 # key: cost type
 # value: cost class
 g_cost_map = {}
diff --git a/python/paddle/trainer_config_helpers/activations.py b/python/paddle/trainer_config_helpers/activations.py
index ad5cdc0a0eb13f7a58e7d89ebfb79d33a63b75d5..2202d0bf96976d5ca694f1417af5da9c31eaa9f0 100644
--- a/python/paddle/trainer_config_helpers/activations.py
+++ b/python/paddle/trainer_config_helpers/activations.py
@@ -23,9 +23,9 @@ __all__ = ["TanhActivation", "SigmoidActivation",
 
 class BaseActivation(object):
     """
-    A mark for activation class. 
+    A mark for activation class.
     Each activation inherit BaseActivation, which has two parameters.
-     
+
     :param name: activation name in paddle config.
     :type name: basestring
     :param support_hppl: True if supported by hppl. HPPL is a library used by paddle
@@ -194,7 +194,7 @@ class SquareActivation(BaseActivation):
 class ExpActivation(BaseActivation):
     """
     Exponential Activation.
-    
+
     .. math::
        f(z) = e^z.
     """
diff --git a/python/paddle/trainer_config_helpers/layers.py b/python/paddle/trainer_config_helpers/layers.py
index 95fbcc3693c89463b8eedf236908daac51e93b0f..10737c90ccfa3c2249d90f34bfa7a1d3d57bf45b 100644
--- a/python/paddle/trainer_config_helpers/layers.py
+++ b/python/paddle/trainer_config_helpers/layers.py
@@ -1766,7 +1766,7 @@ def img_pool_layer(input, pool_size, name=None,
     :type pool_size_y: int|None
     :param num_channels: number of input channel.
     :type num_channels: int
-    :param pool_type: pooling type. MaxPooling or AveragePooling. Default is
+    :param pool_type: pooling type. MaxPooling or AvgPooling. Default is
                       MaxPooling.
     :type pool_type: BasePoolingType
     :param stride: stride width of pooling.