diff --git a/doc/api/v2/config/layer.rst b/doc/api/v2/config/layer.rst
index 8a8774bd696d60945c87487acbd3c2dee5a0c0bd..05817ec85455ac58566e90956a54cb86541f8488 100644
--- a/doc/api/v2/config/layer.rst
+++ b/doc/api/v2/config/layer.rst
@@ -110,7 +110,7 @@ sum_to_one_norm
     :noindex:
     
 cross_channel_norm
----------------
+------------------
 ..  automodule:: paddle.v2.layer
     :members: cross_channel_norm
     :noindex:
diff --git a/paddle/gserver/layers/CrossChannelNormLayer.cpp b/paddle/gserver/layers/CrossChannelNormLayer.cpp
index 0bc90966e2bfe03293175e69353b274d63718580..0c8156ae7736d923bc55df696808fad09bfa1796 100644
--- a/paddle/gserver/layers/CrossChannelNormLayer.cpp
+++ b/paddle/gserver/layers/CrossChannelNormLayer.cpp
@@ -78,7 +78,6 @@ void CrossChannelNormLayer::backward(const UpdateCallback& callback) {
   Matrix::resizeOrCreate(sampleBuffer_, channels_, spatialDim, false, useGpu_);
   scaleDiff_->zeroMem();
   for (size_t i = 0; i < batchSize; i++) {
-    spatialBuffer_->zeroMem();
     // propagate to param.
     MatrixPtr dataBufferTmp =
         Matrix::create(dataBuffer_->getData() + i * dataDim,
diff --git a/paddle/gserver/layers/NormLayer.h b/paddle/gserver/layers/NormLayer.h
index d896abbd757b89694e0cd6ec25adaa100ce4443e..f490f506a9028580f590c013e25d3f54d2ae470c 100644
--- a/paddle/gserver/layers/NormLayer.h
+++ b/paddle/gserver/layers/NormLayer.h
@@ -68,7 +68,7 @@ public:
 /**
  * This layer applys normalization across the channels of each sample to a
  * conv layer's output, and scales the output by a group of trainable factors
- * whose equal to the number of channels.
+ * whose dimensions equal to the number of channels.
  * - Input: One and only one input layer are accepted.
  * - Output: The normalized data of the input data.
  * Reference: